Congratulations to both of you, you and the viewer who helped with the comment. It's just so satisfying to watch you zero in on the issue, fix the problem and bring these old computers back to life. Ohh, and it looks like a nice game, it seems like it's a proper game for the season :)
Your diagnostic capabilities are first rate. I was extremely impressed with your resistor change to brighten the screen and the discovery of the fan causing the interference. Your solution of using the electrolytic cap to filter out the noise was an excellent work around. I noticed there was little frequency change to the clock signal when you were adjusting the variable cap and I thought to myself, that doesn't look right. Then seeing on the schematics that, wide open the chip oscillates at 50 and it was running at 47, it clicked in my head that variable cap was likely the cause. It was obviously running the clock signal wide open. So, I felt good about myself by being suspect of that part and then watching later in the video to find out that my suspicion was correct on that part. Keep up the great work Adrian 🙂
I feel like that would be a perfect April Fools Video. "Welcome to Adrian's Analog Attic, today we will be repairing a carlson-stromberg vacuum tube radio, here in my attic shop..."
That fan could be what was screwing up your disk drive too, not just the display. Nice diag journey. Pro tip: take pictures of every stage of disassembly.
The S symbol on the older PLL chip is for Signetics, which was bought by Philips in 1975, but continued operating as a separate brand. It is the company that invented the 555
Ah Adrian's such a polite chap: "There's a lot of analogue magic going on in this area..." ie: Whomever Tandy hired to 'design' their TRS-80 computers was a bodger extraordinaire! Q: "So how does this bit work?" A: Pfft... "No idea! I just arranged the IC's in neat rows then threw bodge wires and passive components over the top until it did something."
Wow! This takes me back… i had one when i was in the Marines in 1983… 40 years later I still remember all the fiddling i went through with mine. Excellent vid, thanks for sharing
We HAM ops do suffer so much from these plasic trimmer caps dying after years. For example the ICOM IC-471 has over 30 plasic caps in the YGR and other RF units. Unfortunately thes caps sort of depend on each other and there is no service manual that explains how to align them to what... You can only rely on one of two schematics with some pencil notes that some ICOM tech guys made 20 or more years ago. SO thanks to these guys, whoever it was! And thanks Adrian for another entertaining and informative video!
This video is at the top of it's class. Tandy or not, this one was rich with all the good stuff I come to your channel for. You share what you're thinking at every step so naturally, it is as if I was right there with ya, enjoying it all. Now I'll be more inspired to dive into my own Model 4P's. Simply outstanding!
I used to admire the TRS80 in every Tandy store going in just to see them and dreaming of owning one as a kid. I still would! That and the BBC ZX80 and Spectrum later on: all great machines in their time!
2:07 - Whenever I fix anything I always try to write down a description of what I did (especially if it's a mod of some sort) on a piece of paper and leave it either taped inside the device or with the accompanying paperwork. Not only does it help to remind myself what work I did, but the next person doing a repair will certainly appreciate it.
Wow I am impressed of your ability to grasp when you do say that you probably don't... Anyways, here is a crash course in PLL controlled VCO, (commonly used in modern radio receivers and transmitters) Pin 2 is LOOP voltage, it has a legal range, if adjusting the small cap, the LOOP voltage should change, indicating VCO (voltage controlled oscillator) is trying to change frequency, but is forced to stay since the controlling voltage changes, keeping it in Phase locking range. Those caps tend to fail over time, especially when touching them :( they do contain some silver that tend to vanish over time due to oxidation. TIPS: take a couple of single strand leads and twist them hard (do not connect the back end) and you have a variable cap: loosen them gives less cap. tightening them gives more cap. To replace with a fixed value cap that you don't have: Take a standard ceramic disc capacitor an nip the top of it to reduce the capacitance., when done, secure the cap with some nail varnish, you don't want any moist to enter the insides of a small cap :) Good lock and keep up the good work!!!:
Cool to see that thing actually work, as an owner of a TRS-80 model 4p without any disks for it... maybe mine works still, maybe it doesn't, haven't tried in years.
Adrian, you should put a label in every machine you repair with a TH-cam video link, so for future owners who get this machine they can see it featured in your video. That is of course if TH-cam stll exists by then 😉. Thank you for making these videos!
It all seems like magic when you see the interaction of analogue components driving digital hardware, even moreso that they ultimately put all that circuitry inside its' own ICs later on, definitely was quite the boom era when people were churning out new ideas and designs to make computers an affordable and practical thing... :D
Those Arctic F series fans are quite a bad choice for longevity, but they support a feature that isn't advertised for legal reasons (enermax has a patent on it). You can push the fan blades (as in: the entire white rotor assembly) out of the motor to clean them, which also gives you access to the shaft and bearing, meaning you can clean out (most of the) original lubricant and replace it with something that'll last longer than 3-6 months before the bearing runs dry. I've used some (obviously fresh/new) car engine oil filled into a little syringe for the longest time for relubricating fans and it has worked perfectly, unlike light machine oil which just caused things to get noisier somehow. Ideally you would get a dual ball bearing fan and never have to worry about any of that but you know... $€£ ...edit: they also have very low static pressure, so while they move a lot of air freestanding, something like a stamped fan grille will kill off a lot of that flow. just something to keep in mind depending on where they're used
I just bought one of these on eBay and got super lucky. It is not yellow at all, looks like someone stored it for years and it appears to be brand new. I put in the Hi-Res graphics board and the inside was completely clean. I'm still learning how to use it and I do not have a model 3 boot disk yet but I am enjoying using this.
An absolutely fascinating video. I loved how you are really digging into the circuit diagrams and conforming the outputs of each component to find the problem. Bravo!
Amazing videos. It's such a pleasure to see you tinker with these old machines. This was a wonderful tech, problem solving, and hands-on history lesson.
Its great that you kept going when your understanding of the _analogue_ circuits was lacking, and nice you didn't have to go too far to get better understanding (comments and the parts diagram). My brain works in _digital_ by default, so I too understand the consternation that can happen when dealing with certain circuits outside of that _comfort zone_ . It seems (from the video at least) that this machine was suffering from a "bunch" of _small_ problems, which when combined, seemed to create one _large_ problem. I think most people would agree, its nice to see old hardware back in a usable state, even if it takes a few videos to achieve that result. What you could do with that original PLL chip, is solder it into a small test circuit that is designed to help "de-mystify" some of those _analogue_ "mysteries".
Cool. Great to see it working. The trimmer cap (60pf ?) would allow for the exact PLL free running frequency to be set. Removing the jumper is probably for that same purpose so its unlocked while adjusting it. Once its set pretty close (within maybe 1Mhz +/- or better) then replacing the jumper allows for full lock and you're done. I think you hit lucky with the 33pF getting close enough, a trimmer would be more precise. The PLL circuit has a set bandwidth where it can lock. If the free running VCO is too far away from that, it will not lock.
I also suspect that you would be able to tune out the jittering being seen. There's an NE564 used in the Amiga 3000's "display enhancer" circuit and tuning the VC helps stabilise the image
This was great! Loved seeing machines I never really saw or heard very much about. You just might be the leading supplier of that particular PLL chip now. Enjoy running Adrian's PLL Basement!
Nice Star Wars t-shirt! I bought the exact same one on a holiday to Colorado in 2019 (I'm from the UK), I think I bought it in Target. Nice fix on the TRS too 🙂.
Nice job, although I'm cringing a bit at the use of a fixed cap to replace the variable one. Anyway, I wanted to comment on that PLL circuit configuration. The NE564 and 74LS90 combine to make a classic PLL frequency multiplier. You said at one point that the input from the 74LS93 was about 1.2MHz, which makes sense since U148 and U149 act to multiply that frequency by 10 times, producing the 12.5MHz you saw on pin 9 of the NE564 after you got the circuit working. The trick here is that the PLL will try to lock the phase, and thus match the frequency, of the signals on pin 3 and pin 6. The 74LS90 contains two dividers, a divide by 5 from pin 1 to pin 11 and a divide by two from pin 14 to pin 12. Although dividing by 5 produces a non-square wave output, following that with the divide-by-two 2 flip flop squares it up again at 1/10th the initial frequency. This 1/10th frequency is fed back to pin 3 of the PLL. Now when the 1.2MHz signal comes in to pin 6 of the PLL it will try to match the phase/frequency on pin 3, which due to the divide-by-ten function of the 74LS90 it can only do by feeding precisely 12.5MHz into pin 1 of U149. Voila... the PLL output frequency is exactly 10 times that of the output from U147, with the same long term stability as the clock feeding into U147. Note though that short term stability of the PLL output signal is not as solid as the crystal frequency it is derived from, so the PLL VCO must be a stable design with quality components to minimize jitter. PLL frequency multipliers are all around us these days, used in the super accurate frequency synthesizers in cellular telephones, software defined radios, and GPS receivers. You just don't see them now because they are so tightly integrated. The trick with these modern frequency multipliers is that the feedback division can be changed to alter the output frequency digitally. The input frequency to the PLL will be the desired frequency step amount, and the divider will be adjustable to provide a range of outputs frequencies. For instance, an input to the PLL of 50KHz with a divider capable of being switched in increments of two (to preserve the square output from the divider) between 200 and 240 would allow for extremely precise VCO outputs of 10MHz to 12MHz in 100KHz steps. Cheers.
Great explanation! He seemed to luckily hit about the right spot with his fixed cap, and unless circuit parameters deviate too much it should be ok sooo... eh, good enough. Though I'd still recommend ordering and installing a variable cap.
58:38 Did I ever tell you how much I LOVE that vintage '70s green/blue towel majestically draped across the back of your chair? I've seen it in several of your videos, (the AT&T UNIX PC unveiling comes to mind), and I vote you make it your Signature Adrian's Digital Basement vintage computer covering!
This is the first time I've seen 13 Ghosts running in over 35 years. I had a TRS-80 Model I, then a 4P, way back in the day, before I started down the MS-DOS path so long ago.
I made my own desolder braid using copper core speaker or lamp wire. I'd strip and flatten the core and lightly coat one side with flux so it would wick up the solder. Then I'd snip, strip, and repeat the process.
Congrats! Another computer freaking works ;) If the fan was able to upset the video circuitry, it's not unreasonable to assume it did the same with the disk drive hence why it failed reading. Normally this would be a clue on bad caps but.. maybe the designers just didn't put much capacitance on the power rails and it worked anyway? The PLL operates the way you assumed - I'd order a variable cap and install it there if possible, but as long as capacitance is within range it should work fine. The CRT cathode may have degraded even if the computer wasn't running - just from the heat, given how yellowed the computer is. Overvoltaging it may degrade it even faster - although since the computer will probably not be used very often it may be fine. I'm not sure if I'd recommend CRT regeneration but maybe it's worth a try?
You need an old fashioned CRT Restorer, it allows for burning off gun ash and has been used for decades on Picture Tube Televisions. There are also CRT Brighteners, which are transformers that take in the filament voltage and outputs 10 to 15% more AC voltage.
You should be able to run the variable capacitor parallel to a disc cap of a reasonable value and use the variable as a trimmer to get more to the desired value. (It locking in now doesn't mean it will be reliable in the future if there is a different temperature etc.) Many lifetimes ago, I ran a "Towne Crier" BBS on a 4P with the 300 baud "infernal modem" and a pair of 720KB Misubishi floppy drives (80 TK DSDD). IIRC there was a way to add an IC to enable the 4P to run four floppy drives like the Mod 1-4 machines -- the 4P needing a separate ribbon cable for the second two drives. (My Mod 4 ran my store (camera repair wholesale/retail), with four of the half-height drives installed in place of the two full-height drives it came with. I also replaced the switching supply with a big honker of a linear power supply. Why? Because the switching supply burned out (after the warranty ended, of course), and, I had a linear supply that fit inside the cabinet. I eventually sold the 4P, but I still have the Mod4 somewhere (in my Citizen Kane area :) I ran the business on a Mod1, to which I'd added a lowercase mod (a 21l02 chip, I think -- 1Kx1 static), piggybacked onto one of the video RAM chips, with a bunch of jumpers, a pair of surplus calculator keys (used as Ctrl keys), an Assembly language driver written by Dennis Bathory Kitsz (he also wrote the directions for the lowercase mod in his subscription newsetter). Like a fool, I later on added a Mumford Micro turbo kit, and that was the end of my Mod1 (close to four grand, up in smoke, to coin a phrase). Ah, memories...
The video jitter is an artifact of the +12V from the power supply to the video PCB. In your case it was the fan causing problem but often it 'just happens'. I installed a new drop in replacement power supply n my Model 4P from Jay Newirth (Newsoft). It uses a separate +12V supply for the video board and eliminates the problem.
Hi Adrian. Try a 50 pF silver-mica cap instead of the 33 pF ceramic disk cap; the ceramic disk caps aren’t ideal for that application, as their capacitance varies with temperature, so they’ll drift in value as heat builds up in the case. 😢 The silver-mica caps are intended for RF (radio) applications, so should be a lot more stable.
@@therealjammit Its to trim out any error in tolerance within the PLL chip, power supply voltage and anything else that can effect the free running (unlocked) VCO frequency. Once its set close enough, then the PLL can lock and do its magic. Thats why they used a trimmer not a fixed cap. Although in this case, it worked. But in production, maybe not every time.
I would expect a ceramic disc cap with as little as 33 pF to be using some sort of Class 1 dielectric, with little in terms of tempco. Should be plenty good enough for this application. If in doubt, watching the free-running frequency while heating up the cap with hot air should establish how much of an issue it is.
If you wanted, you could put the out-of-spec passive back on the board in parallel with a fixed capacitor. Instead of 33pF exactly, you would want something that adds up to 33 pF when the variable is at the center of its range. Then you'd have the original part still in place, and still providing some measure of tuning, although you're accommodating its operation in a much narrower range than it was specced for.
@@marka1986 If it only continues to get smaller in value, it probably won't matter much because the bulk of the capacitance is in the fixed cap in parallel. It would take deliberate effort to re-form it to increase capacitance, so no worries there. Another option would be to put the part back in place and set it at its absolute minimum and then just pretend it's not there. Leave the 33 pF cap, another 3 to 4 pF probably won't get out of the PLL range. Then as the part ages and becomes completely useless, the total capacitance will actually drift slightly _closer_ to the desired value. :)
Realistically, this is how it should have been done from the factory, but I guess minimum component count won out. You would not normally want to have way excessive adjustment latitude in a finished product. Anyway, it seems 33 pF (luckily) is close enough that he may as well leave that in there and call it a day.
I would also check the caps on the analogue board, if you haven't already. I remember one of the engineers at work telling me one of the PVMs I rescued from the e-waste was trashed because the retrace was visible at normal brightness, which usually indicates the monitor needed a recap. I don't know much about how CRTs work, but I guess it's possible the values or ESR of the caps could also affect the brightness/contrast.
When Adrien is talking about the power supply, he uses the term "REEFA" (?), which I'd never heard used to describe a component. After some searching, I found that he was referring to RIFA, a Swedish manufacturer of resistors and capacitors. The two items he replaces are capacitors, which makes sense, since they are the most likely items to fail with age. And after rewatching the clip several times, I realized he said "capacitor" in reference to them, but I missed it the first few times.
And it's not just because they are capacitors, but because the are across the mains and commonly also before any switch, and when they fail they light on fire. So simply having a thing plugged in and switched off it could still just decide to ignite someday. The newer tech X/Y1/Y2 capacitors no longer ignite so replacing them even if they aren't bad is standard safety procedure. And when they burn they smell horrible and will skunk your house for a week. Similar to Tantalum capacitors, which also love to short when old and also could ignite. Many videos include pre-checks of those orange nasties, but even pre-checks pass and then those short as soon as they get voltage on them. But usually those are on 12v and 5v rails so any sane PSU would shutdown for the short and not just keep shoving amps in, which keeps them from actually lighting on fire, such as the direct-mains Rifa caps, otherwise there would be a policy of just replacing all Tantalums as well.
The low voltage side of the power supply has a pair of resistors in series on a part of the pcb with a square silk screen image. The silk screen might be for a trim pot that has been replaced by the fixed voltage divider, but which would have enabled the fine tuning of the 5v reference voltage.
A lot of TRS-80 Model I/III people loved the 4P even if they never used Model 4 mode, just to have a portable Model I/III system. Also people used them to run BBSes for some reason.
An idea would be to use reallocate-able code. That means once you call it, the entire code runs on relative jumps/calls etc. This would allow you to start testing whatever memory is available with rom mapped in, then select a block that was good to copy the test to. Once copied, it can be called and would identify it is running from ram, and remove rom mapping to test all ram.
The Z80 only has "short" relative jumps and if memory serves me correctly, has no relative calls. It might be possible to make code entirely relocatable by using something similar to the old DOS .EXE format. The file contains both the code and a bitwise table of locations in the program which need to be adjusted, so the loader in the OS reads that table after loading the program and makes the address adjustments within the code before handing control over to the program. It wouldn't be as easy to do on the Z80 though, since Z80 instructions are a mixture of 8-bit and 16-bit instructions, unlike the old 8088/8086 where all instructions were 16-bit.
A flashback to the1980's when I upgraded from a model 1 to a model 4P. I wrote a ton of software on that thing and it even survivde the flood when the water heater in the apartment above me broke and poured water all over the table where the 4P was sitting.
I mentioned in another comment that he could run it parallel with a fixed disc cap of the "appropriate" range, so that he could trim it. But, it's just as possible that the one he dropped in was THE perfect value, so, sometimes, "If it works, don't fix it" ")
@@Colin_Ames -- Well, yeah, it's always best to replace a questionable component with a known-good part, but I was thinking in terms of quickly homing in on the problem by using parts he had on hand. Personally, I'd look to replacing it too, even if I was able to bridge it to bring it back in range. Once it's failing to the extent it's already demonstrated, there's no assurance that it'll _stay_ in its current range rather than continuing to drop to lower and lower capacity.
Looks like one of the inductors on the low voltage side of the power supply might be a variable one? It's got a hex shaped hole in the top, which usually indicates the core can be screwed in or out to vary the inductance. Perhaps that could adjust the voltage of the rails?
The loading of the Model-3 roms reminds me of the Kickstarter disks of the Amiga. Maybe the designers when running low on time borrowed a little something from this machine.
So, you should put the trimmer capacitor back in and then connect the 22pF in parallel. This will allow you to adjust the VCO to maximize the stability of the output clock.
I wonder if the closer you get the free-running VCO to the desired lock frequency, the less hunting / jitter noise you get in the output clock. I don't know how stable the timebase is on your computer 'scope, but it seemed like the frequency was wandering a bit, like vibrato if it was an audio signal. You'd be able to see this on a spectrum analyzer as well.
PLLs seem to hold pretty well once locked on. I remember older radios that started using the PLL would lock onto a station with about the same quality once you were anywhere near the frequency. The SECAM version of the Atari 2600 made a PLL using a VCO chip and a mix of digital and analog components to find the center frequency for color generation. As long as the capacitor-tuned frequency was close enough to the required one that it didn't need excessive voltage on the VCO, the tuning of the capacitor made no visible difference. At the extremes though, it made some interesting artifacts.
I've enjoyed your TRS-80 model 3 and 4P videos as much as anything you have ever done. I like when you work on these big jigsaw puzzles, and following along with your diagnosis is so enjoyable. Only one thing worries me: did you figure out where that small cable attaches to the power supply, or is it still disconnected? What a great fix, with TWO bonuses you pulled out of the hat, viz. improving the CRT brightness and addressing the interference from the new fan. Impressive!
Hi Adrian, when adjusting those tiny trimmer caps, only use a plastic trim tool and they only have a 360 degree adjustment, just very small movements will make the difference.
He did - the cap was just totally bust and probably didn't make good contact internally. I ran into similar issues with variable caps and resistors, some seem to be just too brittle.
maybe you already know this, but when you format a lower density floppy on a drive designed for higher density, it often gives problems because the magnetic head for high density is smaller so it doesnt write the full track. It can read lower density floppies fine but can only write lower density floppies for use in other high density drives.
If I remember my electronics classes (30 years ago already?!) there were two frequencies associated with a PLL circuit: the capture range and the lock range. Without digging out my books, it seems that the input must be within the capture range, and the output is maintained within the lock range. I think.
I often use clamps like seen at 36:58. I confirm they are really useful, and don't cost much. Of course choose a model which stay locked, but I suppose most do. I have a straight one and a curved.
Great! I had a 4P back in 1984. I started out on a TRS80 Model 1 in 1980. No disk drives on the Model 1. I had to load software with a cassette tape player. Primitive.
My first system was a Model III. It had 2 5 1/4" floppy drives. I was so glad I didn't have to mess with cassette tapes for software. I only used my III's cassette port for sound (via Z80 assembly). I first learned to program on my Model III. I had Level III Basic (of course), a Basic compiler of some sort, and a Pascal compiler. I think I may have had a C compiler.
Just wanted to point out that the problems you were having with disks occasionally being unreadable and disks written on one system not always working on another may be because one or more of your floppy drives are slightly out of alignment. Particularly with the older drives, it was actually necessary to manually readjust the alignment on them sometimes. I remember back in the day using specially-written (and rather expensive) "alignment disks" to measure and correctly tune the disk drive alignment on my old TRS-80 drives.. I have no idea whether you could even find one of those anymore, but it is also possible to adjust it reasonably well just using a known reference "good" disk (such as a floppy that was written and sold as part of a commercial product, instead of one somebody wrote in their own disk drive), by slowly tweaking the adjustment one direction until it becomes unreadable, noting the position, turning it the other way until it becomes unreadable, and then setting it midway between the two, etc.
PLL, cool way to increase frequency and lock them to base clock,used them back in the 80's for controlling bit rate on a sound sampler i was designing,clock doubling can be done by creating a pulse on rise and fall of clock.Good video,im wondering if theat trim cap was replaced with the wrong part,before you got to it?,ive never had one go faulty in my 40+ years as a board repairer.
PLLs are useful to change the CPU clock too. CPUs tend to not like being short cycled and the PLL allows the clock to be slid to the new frequency and avoid that issue. Of course some logic will avoid that issue too but it gets a bit complicated when you have to ensure both high and low state of the clock is not short enough to cause the CPU problems. I used a PLL to change the clock on the Z80 in my MIII from stock to around 10MHz (overlocked Z80H) and back as necessary.
Bad trimmer caps are uncommon but not entirely unheard of in FM tuner repair circles. Very likely brand and environment dependent. Those into old radios may also be familiar with "silver mica disease".
with some switching psu circuits, theres a cap, usually electrolytic, in the high voltage primary side, driving the base of the 'chopper' transistor, if it uses a bipolar transistor, (most likely with this), if this is low capacity, it can cause issues, 'round shouldered' /insufficient drive, causing poor output or regulation, screeching noise, overheating of the transistor, and even its failure, in some cases instantly ... also there may be high value resistors within this part of the circuit, if this/they go high, can also cause similar...
Congratulations to both of you, you and the viewer who helped with the comment. It's just so satisfying to watch you zero in on the issue, fix the problem and bring these old computers back to life. Ohh, and it looks like a nice game, it seems like it's a proper game for the season :)
Your diagnostic capabilities are first rate. I was extremely impressed with your resistor change to brighten the screen and the discovery of the fan causing the interference. Your solution of using the electrolytic cap to filter out the noise was an excellent work around. I noticed there was little frequency change to the clock signal when you were adjusting the variable cap and I thought to myself, that doesn't look right. Then seeing on the schematics that, wide open the chip oscillates at 50 and it was running at 47, it clicked in my head that variable cap was likely the cause. It was obviously running the clock signal wide open. So, I felt good about myself by being suspect of that part and then watching later in the video to find out that my suspicion was correct on that part. Keep up the great work Adrian 🙂
Time for a THIRD channel - "Adrian's Analog Attic."
I feel like that would be a perfect April Fools Video. "Welcome to Adrian's Analog Attic, today we will be repairing a carlson-stromberg vacuum tube radio, here in my attic shop..."
Heh --- I think CRTs from the 80s are about as "analog" as I want to get. Hehe. Definitely much prefer the digital side of things!
The theme tune could be on a theremin. I agree that it would make an awesome April fools' episode
Or if you want to go in the other direction, how about Adrian's Quantum Quonset Hut?
@@billesposito3482 At the same time, Paul Carlson will repair a retro computer?
That fan could be what was screwing up your disk drive too, not just the display. Nice diag journey. Pro tip: take pictures of every stage of disassembly.
The S symbol on the older PLL chip is for Signetics, which was bought by Philips in 1975, but continued operating as a separate brand. It is the company that invented the 555
The "triple nickel" timer. Endlessly useful chip.
@@MrJohndoakes Love the blast from the past.. aka TNT (Triple Nickel Timer)
@@MrJohndoakes Yep! And still used in new circuit designs, mostly the CMOS version.
And the NE 571 COMPANDOR...
And the 2650....
All hail the 555. Everything can ne made from them, you just need a bigger room.
I really enjoy watching your approach to troubleshooting - it's very educational. And finding an out-of-spec passive component was icing on the cake!
Ah Adrian's such a polite chap:
"There's a lot of analogue magic going on in this area..." ie: Whomever Tandy hired to 'design' their TRS-80 computers was a bodger extraordinaire!
Q: "So how does this bit work?"
A: Pfft... "No idea! I just arranged the IC's in neat rows then threw bodge wires and passive components over the top until it did something."
I'm glad it turned out to be an easy fix! It's nice to see it running properly.
One can never get sick of TRS-80 videos. Awesome work.
Brings back memories.
Wow! This takes me back… i had one when i was in the Marines in 1983… 40 years later I still remember all the fiddling i went through with mine. Excellent vid, thanks for sharing
We HAM ops do suffer so much from these plasic trimmer caps dying after years. For example the ICOM IC-471 has over 30 plasic caps in the YGR and other RF units. Unfortunately thes caps sort of depend on each other and there is no service manual that explains how to align them to what... You can only rely on one of two schematics with some pencil notes that some ICOM tech guys made 20 or more years ago. SO thanks to these guys, whoever it was! And thanks Adrian for another entertaining and informative video!
Personally I love watching the repair of less common machines. Thanks!
Now all you need is an Adrian's dance party disc utilizing model 4P features!
Made using the TRS-80 "Dancing Demon" program.
This video is at the top of it's class. Tandy or not, this one was rich with all the good stuff I come to your channel for. You share what you're thinking at every step so naturally, it is as if I was right there with ya, enjoying it all. Now I'll be more inspired to dive into my own Model 4P's. Simply outstanding!
I used to admire the TRS80 in every Tandy store going in just to see them and dreaming of owning one as a kid. I still would! That and the BBC ZX80 and Spectrum later on: all great machines in their time!
2:07 - Whenever I fix anything I always try to write down a description of what I did (especially if it's a mod of some sort) on a piece of paper and leave it either taped inside the device or with the accompanying paperwork. Not only does it help to remind myself what work I did, but the next person doing a repair will certainly appreciate it.
Wow I am impressed of your ability to grasp when you do say that you probably don't...
Anyways, here is a crash course in PLL controlled VCO, (commonly used in modern radio receivers and transmitters)
Pin 2 is LOOP voltage, it has a legal range, if adjusting the small cap, the LOOP voltage should change, indicating VCO (voltage controlled oscillator) is trying to change frequency, but is forced to stay since the controlling voltage changes, keeping it in Phase locking range.
Those caps tend to fail over time, especially when touching them :(
they do contain some silver that tend to vanish over time due to oxidation.
TIPS:
take a couple of single strand leads and twist them hard (do not connect the back end) and you have a variable cap: loosen them gives less cap. tightening them gives more cap. To replace with a fixed value cap that you don't have: Take a standard ceramic disc capacitor an nip the top of it to reduce the capacitance., when done, secure the cap with some nail varnish, you don't want any moist to enter the insides of a small cap :)
Good lock and keep up the good work!!!:
41:25 thank you auto-focus, that cable tie is amazingly sharp and clear!
"...release the smoke with some RIFA madness..." I actually LOLed. :)
Adrian almost got thru the line with a straight face too.
Cool to see that thing actually work, as an owner of a TRS-80 model 4p without any disks for it... maybe mine works still, maybe it doesn't, haven't tried in years.
Adrian, you should put a label in every machine you repair with a TH-cam video link, so for future owners who get this machine they can see it featured in your video. That is of course if TH-cam stll exists by then 😉. Thank you for making these videos!
It all seems like magic when you see the interaction of analogue components driving digital hardware, even moreso that they ultimately put all that circuitry inside its' own ICs later on, definitely was quite the boom era when people were churning out new ideas and designs to make computers an affordable and practical thing... :D
Those Arctic F series fans are quite a bad choice for longevity, but they support a feature that isn't advertised for legal reasons (enermax has a patent on it). You can push the fan blades (as in: the entire white rotor assembly) out of the motor to clean them, which also gives you access to the shaft and bearing, meaning you can clean out (most of the) original lubricant and replace it with something that'll last longer than 3-6 months before the bearing runs dry. I've used some (obviously fresh/new) car engine oil filled into a little syringe for the longest time for relubricating fans and it has worked perfectly, unlike light machine oil which just caused things to get noisier somehow. Ideally you would get a dual ball bearing fan and never have to worry about any of that but you know... $€£ ...edit: they also have very low static pressure, so while they move a lot of air freestanding, something like a stamped fan grille will kill off a lot of that flow. just something to keep in mind depending on where they're used
I just bought one of these on eBay and got super lucky. It is not yellow at all, looks like someone stored it for years and it appears to be brand new. I put in the Hi-Res graphics board and the inside was completely clean. I'm still learning how to use it and I do not have a model 3 boot disk yet but I am enjoying using this.
An absolutely fascinating video. I loved how you are really digging into the circuit diagrams and conforming the outputs of each component to find the problem. Bravo!
Amazing videos. It's such a pleasure to see you tinker with these old machines. This was a wonderful tech, problem solving, and hands-on history lesson.
Its great that you kept going when your understanding of the _analogue_ circuits was lacking, and nice you didn't have to go too far to get better understanding (comments and the parts diagram). My brain works in _digital_ by default, so I too understand the consternation that can happen when dealing with certain circuits outside of that _comfort zone_ . It seems (from the video at least) that this machine was suffering from a "bunch" of _small_ problems, which when combined, seemed to create one _large_ problem. I think most people would agree, its nice to see old hardware back in a usable state, even if it takes a few videos to achieve that result. What you could do with that original PLL chip, is solder it into a small test circuit that is designed to help "de-mystify" some of those _analogue_ "mysteries".
Cool. Great to see it working. The trimmer cap (60pf ?) would allow for the exact PLL free running frequency to be set. Removing the jumper is probably for that same purpose so its unlocked while adjusting it. Once its set pretty close (within maybe 1Mhz +/- or better) then replacing the jumper allows for full lock and you're done. I think you hit lucky with the 33pF getting close enough, a trimmer would be more precise. The PLL circuit has a set bandwidth where it can lock. If the free running VCO is too far away from that, it will not lock.
I also suspect that you would be able to tune out the jittering being seen. There's an NE564 used in the Amiga 3000's "display enhancer" circuit and tuning the VC helps stabilise the image
My Model 4 had the same reduced screen size when running Model I/III programs. I remember there being a note in the user's manual about this.
Nice to see that machine come back to life. The 4p was the last trs80 I owned.
This was great! Loved seeing machines I never really saw or heard very much about. You just might be the leading supplier of that particular PLL chip now. Enjoy running Adrian's PLL Basement!
Great job, Adrian. I loved the way you solved the PLL issue. I loved CP/M.
Lovely job, And good call from the commenter on the NE564 PLL issue. Great stuff
The “K” isn’t missing, it’s at the beginning of the next line down. 😌
Little known fact: hemostats can also be used to pinch off blood vessels during surgery.
And Radio Shack used to sell hemostats for other purposes.
@@mikeking7470 -- Puff, puff pass? 🤣
This was great. Watched all 3 vids back-to-back at 2x speed.
Adrian that split screen is awesome def want more in the future!
Nice Star Wars t-shirt! I bought the exact same one on a holiday to Colorado in 2019 (I'm from the UK), I think I bought it in Target. Nice fix on the TRS too 🙂.
I really enjoy your repair videos. This series makes me realize I never want a TRS-80. It looks like they can't do anything. Very unfriendly too.
Nice job, although I'm cringing a bit at the use of a fixed cap to replace the variable one. Anyway, I wanted to comment on that PLL circuit configuration. The NE564 and 74LS90 combine to make a classic PLL frequency multiplier. You said at one point that the input from the 74LS93 was about 1.2MHz, which makes sense since U148 and U149 act to multiply that frequency by 10 times, producing the 12.5MHz you saw on pin 9 of the NE564 after you got the circuit working. The trick here is that the PLL will try to lock the phase, and thus match the frequency, of the signals on pin 3 and pin 6. The 74LS90 contains two dividers, a divide by 5 from pin 1 to pin 11 and a divide by two from pin 14 to pin 12. Although dividing by 5 produces a non-square wave output, following that with the divide-by-two 2 flip flop squares it up again at 1/10th the initial frequency. This 1/10th frequency is fed back to pin 3 of the PLL. Now when the 1.2MHz signal comes in to pin 6 of the PLL it will try to match the phase/frequency on pin 3, which due to the divide-by-ten function of the 74LS90 it can only do by feeding precisely 12.5MHz into pin 1 of U149. Voila... the PLL output frequency is exactly 10 times that of the output from U147, with the same long term stability as the clock feeding into U147. Note though that short term stability of the PLL output signal is not as solid as the crystal frequency it is derived from, so the PLL VCO must be a stable design with quality components to minimize jitter. PLL frequency multipliers are all around us these days, used in the super accurate frequency synthesizers in cellular telephones, software defined radios, and GPS receivers. You just don't see them now because they are so tightly integrated. The trick with these modern frequency multipliers is that the feedback division can be changed to alter the output frequency digitally. The input frequency to the PLL will be the desired frequency step amount, and the divider will be adjustable to provide a range of outputs frequencies. For instance, an input to the PLL of 50KHz with a divider capable of being switched in increments of two (to preserve the square output from the divider) between 200 and 240 would allow for extremely precise VCO outputs of 10MHz to 12MHz in 100KHz steps. Cheers.
Great explanation!
He seemed to luckily hit about the right spot with his fixed cap, and unless circuit parameters deviate too much it should be ok sooo... eh, good enough. Though I'd still recommend ordering and installing a variable cap.
Oh man this brings back nostalgia for me late 70s early 80s…
58:38 Did I ever tell you how much I LOVE that vintage '70s green/blue towel majestically draped across the back of your chair? I've seen it in several of your videos, (the AT&T UNIX PC unveiling comes to mind), and I vote you make it your Signature Adrian's Digital Basement vintage computer covering!
This is the first time I've seen 13 Ghosts running in over 35 years. I had a TRS-80 Model I, then a 4P, way back in the day, before I started down the MS-DOS path so long ago.
I made my own desolder braid using copper core speaker or lamp wire. I'd strip and flatten the core and lightly coat one side with flux so it would wick up the solder. Then I'd snip, strip, and repeat the process.
Congrats! Another computer freaking works ;)
If the fan was able to upset the video circuitry, it's not unreasonable to assume it did the same with the disk drive hence why it failed reading.
Normally this would be a clue on bad caps but.. maybe the designers just didn't put much capacitance on the power rails and it worked anyway?
The PLL operates the way you assumed - I'd order a variable cap and install it there if possible, but as long as capacitance is within range it should work fine.
The CRT cathode may have degraded even if the computer wasn't running - just from the heat, given how yellowed the computer is. Overvoltaging it may degrade it even faster - although since the computer will probably not be used very often it may be fine. I'm not sure if I'd recommend CRT regeneration but maybe it's worth a try?
Great series - the Tandy videos are great fun. Well done Adrian!
Adrian, your videos are so satisfying! Thanks for concluding with solutions!
Really enjoyable to see your troubleshooting process. Primes my brain with ideas when I go to fix things.
track down a trimmer cap and replace that 33pf. You never know when after some time it will be unable to Phase Lock again. Awesome video !
I own one of these glorious machines, and a model I level II. Outstanding 😁
You need an old fashioned CRT Restorer, it allows for burning off gun ash and has been used for decades on Picture Tube Televisions. There are also CRT Brighteners, which are transformers that take in the filament voltage and outputs 10 to 15% more AC voltage.
Medical forceps and hemostats are fantastic tools to have around for a variety of applications.
You should be able to run the variable capacitor parallel to a disc cap of a reasonable value and use the variable as a trimmer to get more to the desired value. (It locking in now doesn't mean it will be reliable in the future if there is a different temperature etc.)
Many lifetimes ago, I ran a "Towne Crier" BBS on a 4P with the 300 baud "infernal modem" and a pair of 720KB Misubishi floppy drives (80 TK DSDD).
IIRC there was a way to add an IC to enable the 4P to run four floppy drives like the Mod 1-4 machines -- the 4P needing a separate ribbon cable for the second two drives. (My Mod 4 ran my store (camera repair wholesale/retail), with four of the half-height drives installed in place of the two full-height drives it came with.
I also replaced the switching supply with a big honker of a linear power supply. Why? Because the switching supply burned out (after the warranty ended, of course), and, I had a linear supply that fit inside the cabinet.
I eventually sold the 4P, but I still have the Mod4 somewhere (in my Citizen Kane area :)
I ran the business on a Mod1, to which I'd added a lowercase mod (a 21l02 chip, I think -- 1Kx1 static), piggybacked onto one of the video RAM chips, with a bunch of jumpers, a pair of surplus calculator keys (used as Ctrl keys), an Assembly language driver written by Dennis Bathory Kitsz (he also wrote the directions for the lowercase mod in his subscription newsetter).
Like a fool, I later on added a Mumford Micro turbo kit, and that was the end of my Mod1 (close to four grand, up in smoke, to coin a phrase).
Ah, memories...
Wonderful episode. Was fun seeing CP/M too. Back in the day Pip and I were good friends, even if it was a bit irksome ;)
The video jitter is an artifact of the +12V from the power supply to the video PCB. In your case it was the fan causing problem but often it 'just happens'. I installed a new drop in replacement power supply n my Model 4P from Jay Newirth (Newsoft). It uses a separate +12V supply for the video board and eliminates the problem.
It was also causing the CRC error. The drive power was inconsistent causing speed variations on the motor.
Hi Adrian. Try a 50 pF silver-mica cap instead of the 33 pF ceramic disk cap; the ceramic disk caps aren’t ideal for that application, as their capacitance varies with temperature, so they’ll drift in value as heat builds up in the case. 😢 The silver-mica caps are intended for RF (radio) applications, so should be a lot more stable.
Phase locked loop. I don't know why they used a variable capacitor in the first place. The PLL will lock on if the free run frequency is close enough.
@@therealjammit Its to trim out any error in tolerance within the PLL chip, power supply voltage and anything else that can effect the free running (unlocked) VCO frequency. Once its set close enough, then the PLL can lock and do its magic. Thats why they used a trimmer not a fixed cap. Although in this case, it worked. But in production, maybe not every time.
I would expect a ceramic disc cap with as little as 33 pF to be using some sort of Class 1 dielectric, with little in terms of tempco. Should be plenty good enough for this application. If in doubt, watching the free-running frequency while heating up the cap with hot air should establish how much of an issue it is.
you could use a metal tube as spacers that would work if you wanted to put in a different crt with the tab mountings in a different location
Is done is good .. really liked the gremlins hunt on this one .. and it does look like a cool little machine definitely worth preserving..
If you wanted, you could put the out-of-spec passive back on the board in parallel with a fixed capacitor. Instead of 33pF exactly, you would want something that adds up to 33 pF when the variable is at the center of its range. Then you'd have the original part still in place, and still providing some measure of tuning, although you're accommodating its operation in a much narrower range than it was specced for.
Good idea in general, but I wouldn't trust that trimmer not to drift.
@@marka1986 If it only continues to get smaller in value, it probably won't matter much because the bulk of the capacitance is in the fixed cap in parallel. It would take deliberate effort to re-form it to increase capacitance, so no worries there.
Another option would be to put the part back in place and set it at its absolute minimum and then just pretend it's not there. Leave the 33 pF cap, another 3 to 4 pF probably won't get out of the PLL range. Then as the part ages and becomes completely useless, the total capacitance will actually drift slightly _closer_ to the desired value. :)
Realistically, this is how it should have been done from the factory, but I guess minimum component count won out. You would not normally want to have way excessive adjustment latitude in a finished product. Anyway, it seems 33 pF (luckily) is close enough that he may as well leave that in there and call it a day.
I have learned SO MUCH from these 4P vids. I am stoked!
Nice! Had to watch this in 2 parts because of the length but it's freeking awesome!
Pro Tip: To quieten a piezo speaker/beeper a blob of blu tack works wonders! Semi permanent and easy to reverse 😉
Heh! True. I've actually found just some tape also works super well.
I would also check the caps on the analogue board, if you haven't already. I remember one of the engineers at work telling me one of the PVMs I rescued from the e-waste was trashed because the retrace was visible at normal brightness, which usually indicates the monitor needed a recap. I don't know much about how CRTs work, but I guess it's possible the values or ESR of the caps could also affect the brightness/contrast.
When Adrien is talking about the power supply, he uses the term "REEFA" (?), which I'd never heard used to describe a component. After some searching, I found that he was referring to RIFA, a Swedish manufacturer of resistors and capacitors. The two items he replaces are capacitors, which makes sense, since they are the most likely items to fail with age. And after rewatching the clip several times, I realized he said "capacitor" in reference to them, but I missed it the first few times.
We always called them Y-caps as he mentioned. Non-polarized and used in power supplies I believe to get rid of hi frequency ripple.
And it's not just because they are capacitors, but because the are across the mains and commonly also before any switch, and when they fail they light on fire. So simply having a thing plugged in and switched off it could still just decide to ignite someday. The newer tech X/Y1/Y2 capacitors no longer ignite so replacing them even if they aren't bad is standard safety procedure. And when they burn they smell horrible and will skunk your house for a week.
Similar to Tantalum capacitors, which also love to short when old and also could ignite. Many videos include pre-checks of those orange nasties, but even pre-checks pass and then those short as soon as they get voltage on them. But usually those are on 12v and 5v rails so any sane PSU would shutdown for the short and not just keep shoving amps in, which keeps them from actually lighting on fire, such as the direct-mains Rifa caps, otherwise there would be a policy of just replacing all Tantalums as well.
The low voltage side of the power supply has a pair of resistors in series on a part of the pcb with a square silk screen image. The silk screen might be for a trim pot that has been replaced by the fixed voltage divider, but which would have enabled the fine tuning of the 5v reference voltage.
Is fun to watch you bring some of our computer history back to life,, What you do definitely beats sitting at a bar or tavern somewhere,, :-)
Hemostats are also great as heat sinks when soldering/desoldering.
A lot of TRS-80 Model I/III people loved the 4P even if they never used Model 4 mode, just to have a portable Model I/III system. Also people used them to run BBSes for some reason.
An idea would be to use reallocate-able code. That means once you call it, the entire code runs on relative jumps/calls etc.
This would allow you to start testing whatever memory is available with rom mapped in, then select a block that was good to copy the test to. Once copied, it can be called and would identify it is running from ram, and remove rom mapping to test all ram.
The Z80 only has "short" relative jumps and if memory serves me correctly, has no relative calls. It might be possible to make code entirely relocatable by using something similar to the old DOS .EXE format. The file contains both the code and a bitwise table of locations in the program which need to be adjusted, so the loader in the OS reads that table after loading the program and makes the address adjustments within the code before handing control over to the program.
It wouldn't be as easy to do on the Z80 though, since Z80 instructions are a mixture of 8-bit and 16-bit instructions, unlike the old 8088/8086 where all instructions were 16-bit.
A flashback to the1980's when I upgraded from a model 1 to a model 4P.
I wrote a ton of software on that thing and it even survivde the flood when the water heater in the apartment above me broke and poured water all over the table where the 4P was sitting.
Nice work, as always. Personally I would replace the fixed capacitor with the correct value trimmer, but that’s just me.
I mentioned in another comment that he could run it parallel with a fixed disc cap of the "appropriate" range, so that he could trim it. But, it's just as possible that the one he dropped in was THE perfect value, so, sometimes, "If it works, don't fix it" ")
@@This_is_my_real_name There’s certainly nothing wrong with that approach, I was just saying what I would do. Each to their own.
@@Colin_Ames -- Well, yeah, it's always best to replace a questionable component with a known-good part, but I was thinking in terms of quickly homing in on the problem by using parts he had on hand. Personally, I'd look to replacing it too, even if I was able to bridge it to bring it back in range. Once it's failing to the extent it's already demonstrated, there's no assurance that it'll _stay_ in its current range rather than continuing to drop to lower and lower capacity.
Looks like one of the inductors on the low voltage side of the power supply might be a variable one? It's got a hex shaped hole in the top, which usually indicates the core can be screwed in or out to vary the inductance. Perhaps that could adjust the voltage of the rails?
“Alright, here we are with some hot split screen action “ broke me! 🤣🤣
Congrats, well done, really enjoyed this one..
Would you be kind enough to share your 80 col RGB2HDMI profile please
The loading of the Model-3 roms reminds me of the Kickstarter disks of the Amiga. Maybe the designers when running low on time borrowed a little something from this machine.
Colleagues on Monday morning: "Hey Catherine what did you do on the weekend? - Hot splitscreen action with Adrian."
So, you should put the trimmer capacitor back in and then connect the 22pF in parallel. This will allow you to adjust the VCO to maximize the stability of the output clock.
I wonder if the closer you get the free-running VCO to the desired lock frequency, the less hunting / jitter noise you get in the output clock. I don't know how stable the timebase is on your computer 'scope, but it seemed like the frequency was wandering a bit, like vibrato if it was an audio signal. You'd be able to see this on a spectrum analyzer as well.
PLLs seem to hold pretty well once locked on. I remember older radios that started using the PLL would lock onto a station with about the same quality once you were anywhere near the frequency.
The SECAM version of the Atari 2600 made a PLL using a VCO chip and a mix of digital and analog components to find the center frequency for color generation. As long as the capacitor-tuned frequency was close enough to the required one that it didn't need excessive voltage on the VCO, the tuning of the capacitor made no visible difference. At the extremes though, it made some interesting artifacts.
I've enjoyed your TRS-80 model 3 and 4P videos as much as anything you have ever done. I like when you work on these big jigsaw puzzles, and following along with your diagnosis is so enjoyable. Only one thing worries me: did you figure out where that small cable attaches to the power supply, or is it still disconnected? What a great fix, with TWO bonuses you pulled out of the hat, viz. improving the CRT brightness and addressing the interference from the new fan. Impressive!
Neatly done! Don't ya hate it when the component you suspect is bad turns out to be good? (Makes ya turn around and think a bit...)
Perhaps the fan noise was also causing the CRC errors.
What an adventure! Love it!
Nice job getting it up and running.
Hi Adrian, when adjusting those tiny trimmer caps, only use a plastic trim tool and they only have a 360 degree adjustment, just very small movements will make the difference.
He did - the cap was just totally bust and probably didn't make good contact internally.
I ran into similar issues with variable caps and resistors, some seem to be just too brittle.
Maybe this fan had an effect on the drive too.
Maybe the interference from the fan screwed up reading the disk drive too?
Exactly what I thought.
Yes 12V noise caused motor speed variations in the drive.
maybe you already know this, but when you format a lower density floppy on a drive designed for higher density, it often gives problems because the magnetic head for high density is smaller so it doesnt write the full track. It can read lower density floppies fine but can only write lower density floppies for use in other high density drives.
If I remember my electronics classes (30 years ago already?!) there were two frequencies associated with a PLL circuit: the capture range and the lock range. Without digging out my books, it seems that the input must be within the capture range, and the output is maintained within the lock range. I think.
I often use clamps like seen at 36:58.
I confirm they are really useful, and don't cost much.
Of course choose a model which stay locked, but I suppose most do.
I have a straight one and a curved.
Wow. You are simply the best!
Haven't seen CP/M Plus on a 4p, just a Model-12 or 16. Haven't seen the TRSDOS bootup screen of a computer in years, awesome.
Never underestimate the power of broken passives.
Great video as usual!
Great! I had a 4P back in 1984. I started out on a TRS80 Model 1 in 1980. No disk drives on the Model 1. I had to load software with a cassette tape player. Primitive.
My first system was a Model III. It had 2 5 1/4" floppy drives.
I was so glad I didn't have to mess with cassette tapes for software. I only used my III's cassette port for sound (via Z80 assembly).
I first learned to program on my Model III. I had Level III Basic (of course), a Basic compiler of some sort, and a Pascal compiler. I think I may have had a C compiler.
Just wanted to point out that the problems you were having with disks occasionally being unreadable and disks written on one system not always working on another may be because one or more of your floppy drives are slightly out of alignment. Particularly with the older drives, it was actually necessary to manually readjust the alignment on them sometimes.
I remember back in the day using specially-written (and rather expensive) "alignment disks" to measure and correctly tune the disk drive alignment on my old TRS-80 drives.. I have no idea whether you could even find one of those anymore, but it is also possible to adjust it reasonably well just using a known reference "good" disk (such as a floppy that was written and sold as part of a commercial product, instead of one somebody wrote in their own disk drive), by slowly tweaking the adjustment one direction until it becomes unreadable, noting the position, turning it the other way until it becomes unreadable, and then setting it midway between the two, etc.
PLL, cool way to increase frequency and lock them to base clock,used them back in the 80's for controlling bit rate on a sound sampler i was designing,clock doubling can be done by creating a pulse on rise and fall of clock.Good video,im wondering if theat trim cap was replaced with the wrong part,before you got to it?,ive never had one go faulty in my 40+ years as a board repairer.
PLLs are useful to change the CPU clock too. CPUs tend to not like being short cycled and the PLL allows the clock to be slid to the new frequency and avoid that issue. Of course some logic will avoid that issue too but it gets a bit complicated when you have to ensure both high and low state of the clock is not short enough to cause the CPU problems. I used a PLL to change the clock on the Z80 in my MIII from stock to around 10MHz (overlocked Z80H) and back as necessary.
Bad trimmer caps are uncommon but not entirely unheard of in FM tuner repair circles. Very likely brand and environment dependent. Those into old radios may also be familiar with "silver mica disease".
It's funny to see you talk like that about the fan, Arctic makes some of the quietest and most efficient fans on the market today.
OMG the old Trash 80s... Was the very first computer I touched as a kid back in the early 80s.... Good times.
Any reason October couldn't be Octandy? ;) Would especially like to see a demo of that TRS-80 Micro that you teased recently...
with some switching psu circuits, theres a cap, usually electrolytic, in the high voltage primary side, driving the base of the 'chopper' transistor, if it uses a bipolar transistor, (most likely with this), if this is low capacity, it can cause issues, 'round shouldered' /insufficient drive, causing poor output or regulation, screeching noise, overheating of the transistor, and even its failure, in some cases instantly ... also there may be high value resistors within this part of the circuit, if this/they go high, can also cause similar...