HP 4957A Serial Protocol Analyzer
ฝัง
- เผยแพร่เมื่อ 26 ก.ย. 2024
- We make two good HP 4957A out of two bad ones.
ROM dumps and other info here:
www.curiousmar...
Our sponsors
- PCBWay: fast turn PCBs, www.pcbway.com
- Keysight: test instruments: www.keysight.com
- Samtec: connectors: www.samtec.com
Support the team on Patreon: / curiousmarc
Buy shirts on Teespring: teespring.com/...
Learn more on the companion site: www.curiousmar...
Contact info: www.youtube.co...
I used the prior generation of this product in 1985 when I was working for Raytheon as a High School intern (!). We were going SNA over SDLC and BiSync - and used this to monitor SNA handshakes to a 3270 product for the airlines. It brings back the Memories!
Syn syn syn enq!😁
When I worked as a Systems Engineer at HP in the 80s, I would sometimes use a version of this to troubleshoot mis-wired and mis-configured serial connections. It made the job a breeze. Loved it!
1:22 The Why cable. Is that the one used when shouting at the circuit going why!? why are you not working? ;)
Some interesting HP trivia at 11:22. The power supply has “Yokogawa” printed at the upper left. Yokogawa was a Japanese company that HP had a joint venture with for many years. Lots of products came out of “YHP” as it was called. I wonder if just the PS came from YHP or if the entire 4957A was designed and manufactured there. I don’t remember. Maybe someone else knows.
AFAIK Yokogawa still exists
Hi Marc, I am incredibly jealous - I bought a faulty one off e-bay hoping it was a power supply issue - well it was but the power supply was blown because the RS-232 port had been hooked to mains! That pretty much blew out everything on the 12V lines - so the floppy drive is dead (stepper drive blown) the power supply had the same fault with the start cap as yours, but when I fixed that the PSU exploded. The vertical scan was blown and the RS-232 line drivers and interface board were destroyed. Note - a video will come!
In debugging the system I have found a lot - the video is handled using a dual-port RAM from IDT - this is big enough for the 80 column mode - on the machine that is playing up in 80 column mode, it could be worth having a look at the input and output address lines and data lines - that could give clues as to where the fault is. It may just be the IDT RAM has an issue - in which case it is a simple switch out is possible as it is a PLCC soldered onto the board.
Mine appears to be an early version - it does not have the 80 column VT100 emulator firmware. So maybe the firmware was rushed out to get the unit in the field - makes me wonder if there was a serious supply problem with the older version.
Oh no! Looks like I took all the eBay good luck from you! Still took me two tries. We should definitely try the simple RAM replacement if it's socketed.
@@CuriousMarc not socketed but not fine pitch
mains on RS-232 ? sometimes you really wonder wtf people do with their gear. When I worked for a company that created DMX-512 (RS-485 based) lighting controllers we once got one returned for repair where we discovered the varistor that was protecting the DMX-512 output was just... gone, a scorched crater left in its place. o.O
@@MatthijsvanDuin It is the only way to explain the level and nature of the damage. It is amazing the number of people who should know better but don't. A few years ago a customer returned a sub-station control computer complaining it was faulty out of the box. When we opened it every board was either burnt or was smoke damaged. Somebody connected it to 380V not 230V. Varistors do burn well and smell bad - and when you have 20 or 30 of then all overloaded hilarity ensues!
I see. The version we used was the earlier one with the breakout box in the cover. Great help while diagnosing problems.
I must have resoldered hundreds of those power supply pins that cracked. The first ones cracked again, so I changed to cleaning all the solder away. Then I wound an inch or so of 24 AWG bare copper wire around the pin with a tail running into the PCB trace. Never had an intermittent problem again after that. 👍👍
Thanks for another great video, Marc. Stay curious!
@9:38 - "...as a historical side note..." We come to expect nothing less, fine sir.
Went through many airport check-ins with on of these as a carry on. Once, I had to demonstrate that it powered up to security, and it fell all the way to the floor. Still worked!
It was decades ago but I used one of these to troubleshoot RS-232 communications for test equipment I designed for production test of aircraft flight data acquisition units. It really helped.
Nice work as always! Interesting with the ROM swap.
The starting cap I had to replace in a PSU recently too and learned about that, nice with older switched PSU's because there simple enough to visually figure out what they do.
Funny, I had to replace this capacitor in a terminal power supply recently. In my case and in the video, they are both mounted next to hot components. The failures were almost certainly caused by heat.
Enjoyed the video! I have a 4952 and used it to debug more serial protocols than I can remember. Super handy tool.
Magic fingers, beeps of joy, and working electronics! It's what Saturday mornings are made for 😊 Oh, and sofas & coffee 😉
I used to work with a flavour of this a lot back in the '80's. Ours had a kind of pod thing that covered the keyboard when it was folded up and it then became part of the case. Once removed the pod could be velcroed to the top pouch, attached to the rear panel with a multipin connector and the pod presented the breakout above the screen. Never seen one with the BOB on the side. Very expensive bit of kit back in the day.
Gosh I would have loved to have had something like this back in my ISP days in the 1990s, would have made my job so much easier.
Have one of these and still use it a lot now - many industrial machines still use 232 or 485 etc - Great piece of kit :)
Great work! I've got a pair of the 4952A's that I picked up to do Bit Error Rate testing on old analog modem circuits or their modern digital equivalent (Term Server - Cisco Router - T-1 Line - Cisco Router - Term Server). You would think that the digital equivalent would just work but there are various T-1 timing options that can be misconfigured in the routers or by the T-1 provider in ways that can result in mangled data just like we used to see on old analog modems.
Glad to see I'm not the only one battling obscure T1 timing issues that nobody seems to understand anymore
@@TheFool2cool My low-level headache from my flu shot finally subsided so I thought I'd punch up a quick YT video.
Now I'm mired in horrible T1 clocking memories (internal, line, loopback tail circuits) and my headache's back. 🤦
Dedicated leased 4-wire analog multi-drops! DO NOT send a loopback tone to trouble shoot or you'll isolate a half dozen sales offices!
Yeah! Used this in the 1980s to debug X25 and other proprietary protocol implementations I was doing on a PDP11/RSX11 edit: It must’ve been the earlier version as it was around 1983ish
I saw this unit in one of the central offices I visited. It was connected to the rack, I think to control and monitor the old tel exchange or data circuits to customers. This was a neat video for me to see, thank you. These are really powerful units for their time.
Never go full electroboom!
Got HP 4951C for review 😀
With all cables, Floppy disk with utilities and connection board of cource. And yep - It's definitely cute
Good luck with it! The 4951 with its traditional logic is a lot more hackable than the 57 with all the FPGAs!
@@CuriousMarc Yep. but after some research I fount that binary format for apps is incompatible with 4952 and later models. So I could succesfully upload apps from utility disk for 4951, but still can't build my own - Need to have some sex with IDA Pro to find out entry points addresses.
In the 1980s, each time you needed to connect a new pair of computer and peripheral device like a terminal or printer via a serial interface, the game started over form new. Our rule was: you need a solder iron, an oscilloscope, and an hour of time. We had additional gadgets like back-to-back soldered D-Sub connectors with blank wires in between to hook up the probes. There were so called "null-modem" cables, crossed cables (crossing TX/RX and possibly also DTR/RTS), and all kind of gender changers. And of course there were the 25-pin and space saving 9-pin versions of the D-Sub connectors. What a mess. It was easier if the devices supported the Xon/Xoff flow control, then you could get away with just three wires, you only needed to get TX/RX correctly connected.
Not yet 20 years ago, I designed a test adapter for SCADA systems that ran serial protocols over optical fibers. The adapter consisted of two optical/RS-232 converters which could be combined in tap mode to sniff the communication between the real devices, or the converters could be used independently for simulating either end device in such a system. The processing was no longer done in such a device like the hp 4957A, but with a software running on a PC. The product has been discontinued years ago, but we still get occasional requests for the test adapter hardware.
Well said. RS-232 in a nutshell. No matter how much I try, Rx and Tx are always inverted the wrong way, I end up with two males facing each other, one is 9 pin and the other 25, and hardware handshake freezes the whole thing :-D
@@CuriousMarc It really wasn't that hard if you worked with it a lot, you just had to know the signaling and the ways it was commonly used. I did a lot of serial troubleshooting back in the day. A few minutes with a breakout box and you knew exactly what pinout you needed for any cable. I also used one of these protocol analyzers to decode x.25, baudot and bisync on both async and sync circuits for many years.
This takes me back. This was something we had at work back in the 80s and used to use all the time.
Loved that box back in the day.
Ah the 4957, solved many problems with one of these. Was my go-to tool for anything serial.
Ah the HP4951. Brought back lots of good memories. Used it extensively during the 90's to develop and debug controlling machines for assembly lines. The external breakout panel that you mentioned is actually the front cover, once the keyboard is folded up. It was only in the later models that they integrated the breakout panel into the chassis, like this HP4957.
Congratulations
I have a love for the old HP equipment, made by engineers for engineers. Sometimes even barebones, but every function had a use and every use had a function. While e.g. scopes from Tektronics were almost overly elaborate, HP kept to its swim lane and made excellent tools for the everyday use. Not too keen on the Agilent days, but as Keysight they seem to have found their focus again.
Awesome video as ever, Marc. Those yellowed spacebars are bad for my OCD, though... I'd be happy to retr0bright them for you!
This is a fascinating video. I always learn something new when watching your presentations! Thanks!
It looks good. It beeps. It's beautiful. I want it.
Imagine a channel where its just these 3 guys literally fixing the practically impossible to fix. The channel name could be "Practically Impossible Fixes"
Used these daily (4952a first) for nearly a decade and eventually ended up just looking at the flashing lights to find errors. Lots of repetitive transactions which is why flashing lights were useful. Lived in nearly all of these protocols.
Seemed complex then but today is very simple compared to what kids experience nowadays.
Oh wow. This beings back serial debugging memories.
It’s a wonderful thing when you buy a parts machine to try and make 1 good unit, when the parts machine only needed minor repairs and you end up with 2 good units!
This was great, I didn't realize that these had so many processors either. I have a couple of the 4952A models that I rescued a number of years back from being thrown away that I've never used. They don't have the cable sets and are the model Marc referred to that has the rather clunky interface connections on the lid vs the side. Who know why they made such a drastic change. I'd love to see them working again someday just for fun. It's not too often that [old] protocol decoding for serial communications needs to be done but I hate to see stuff like that abandoned and thrown away. This stuff was great and back when HP made some of the best test equipment in the world.
We used these in the 90s to debug the serial devices that car manufacturers required our computers to talk to. Believe it or not, sometimes the specifications they gave us weren't exactly correct.
Alright, another piece of equipment to go on my ever-growing wishlist... Love these repair videos.
A marvelous toy I last used about 40 years ago!
The first thing I always try to do with those old machines is to look up if Noctua has a matching super silent industrial fan and replace them.
Same. Although it should be said, these quieter fans rarely move the same amount of air - more air per dB of noise, sure, but still less overall airflow. In most cases this is not an issue at all, but I've had it bite me in the ass a few times, where the new quiet fan wasn't quite able to do the job.
11:49 Master Ken wants you to stay alive to do more awesome cooperations :)
The character overlap may be caused by a degraded capacitor in the horizontal width circuit on the CRT. There may be a width coil you can adjust, but by making the image wider (like it originally was), I doubt you'd get the overlap. Looks like an analog issue, not digital.
If you're interested in alleviating your lack of spot welding equipment, the cheapest reliable/quality solution I've found is the kWeld. Paired with a commodity high-current RC hobby lithium ion pouch battery it produces reliable welds and is tunable from the very lightest terminals/strips up to pretty beefy stuff. Not dirt cheap (a couple hundred bucks plus the battery to drive it), but I tried other cheap aliexpress specials before and they blew up in my face on the first weld attempt so don't bother with anything lower-end.
I am pretty certain our electric utility used two of these with our SCADA supervisory control and data acquisition RTU remote telemetry units comm lines.
I would be nice if hp and other manufacturers released schematics and service information, at least for these old pieces of equipment
Oh wow, I would have loved to have this for a project I just finished up! Seems very useful
You can run diffs between binaries using Ghidra
wonderful video, we have a HP 4951A in our lab. that is still good if it is needed for a task.. thanks a lot for the interesting walk around...:) 😊 👍
When your parts machine winds up being better than your main machine. LOL!
I cut my engineering teeth writing and debugging Z8530 SCC drivers and an HP4951C was always at hand to help. I have an HP4952A which I bought out of nostalgia.
It even has EBCDIC. Nice thing to have.
I was not expecting FPGA in such old gear.
It's a good machine. I have one and last "used in anger" as recently as last year for working on a medical laser of similar vintage... modern USB-RS232 dongles didn't work with it. (Eventually we found a fancy industrial one that did, but the HP serial protocol analyzer was still very useful figuring out the settings and some of the protocol.)
I used one of these back in mid 90's to trouble-shoot X.25 traffic blast for the past...
Oh, what better to do at 2:20EST!
I have the exact same VT100 here. I believe it has a 8080cpu.
As always, lovely job. They have fancy floppy drives, I haven't seen ones with cutouts there before.
Perfect. Just take what you get with the given hardware. Plus some recapping to get it working at all. Thanks for sharing ! :-)
Marc, try ESC # 5 and ESC # 6 escape codes on these machines. The first is "single width line" (80 characters) and the second is "double width line" (40 characters). They may or may not do anything useful. There is also an escape sequence for a 132 character line.
That does not do anything on mine.
Oh my, this is such a cutie. I want to adopt one 😀 I have no idea what it does, but nevermind. :)😀
Could have used one of these at work last week. Was trying to troubleshoot a communication problem to a lab machine. Eventually determined that when the PC's power supply failed, it apparently fried the UART on the motherboard.
I got the NavTel 9460 Plus in brand new condition. Those are neat units, this one is RISC based and got an Harddisk built right in too.
OMG haven't seen one of these since I sent like 50 of those to Dove Bid about 15 years ago when I worked for Intel. ;-)
I used these analyzers early in my career for Async, Bisync, SDLC and HDLC analysis. However, I always preferred the Atlantic Research Interview Line of products over the HP analyzers. I spent many years in the office and in the field doing debugging, mostly with the AR Interview 932 model. At least for me, the AR units were simpler and easier to use that the HP ones.
I have one of these, but it looks a little different. It doesn't have the interface on the side. It was a separate piece that acted as a cover over the folded up keyboard and contained the breakout interface and connected to the main unit with a large DB cable. DB37 maybe? It was instrumental in product development that allowed me to start my own company in the early 2000s. Edit: Ahh looks like miine is a 4952..not a 4957.
You have the earlier HP 4951 or HP 4952 that I talk about in the video. Nice machines too!
cool machine, I travelled with that kind of device aroun europe well before laptops. used that as a terminal and also to debug protocols. Used also texas silent, but with bubble memory 😁 and integrated 300b modem... that protocol analyzer phased out later on because it has quite low max baudrate....
machine I used was 4951 (19200b)
yes, new video. :) greetings from slovakia
Greetings from California!
Ah, serial comms.....happy days(!). Used to work at a place that did training and was asked to make a wee box to spit out "The quick brown fox...." endlessly to torture students. Used what was at the time a pretty neat chip, the TMS7782. Used the same chip to make a tape position counter for reel-to-reel studio recorders that output in Braille...but I digress.
I would make software changes to a modem and sweat the BER Bit Error Rate results with the 511 pseudo random sequence. It was a harsh and unfeeling but fair judge of my software effort.
That's great. Now you have a terminal to connect to the high resolution machine. Win win
Finally a good portable Rpi4 case!
I like pretending I'm one of the crowd ... like flying a cardboard rocket ship when you're 10. Wow that was a long time ago.
To me the battery seems just 4S1P (unlike 14:24 audio says a serio-parallel), and the vertical joints extend to a solder pin purely for mechanical reasons
Could very well be.
Verry interesting stuff
I have one of these, and it also has the lower resolution mode for the vt100 emulation, which has always bummed me out since i wanted to use it as a cool looking terminal.
Wonder if this would be of any use for troubleshooting HDLC interfaces on NEMA TS2 traffic signal controllers. They use HDLC to interface to the loop detectors, conflict monitor/malfunction management unit, and the load switches for the signal heads.
Very nice result!
Reminded me of my repair of a 4952A. (Does the 4957 also have this stupid red keyboard connector?)
Looking forward to Eric's reverse engineering results. ;-)
It is a ribbon cable onto a standard DIL header
I could have used something like this a couple years ago at work, believe it or not. Was bringing up a new board, but wasn't getting output. Wasn't until I pulled out an oscilloscope that I found the baud was 33% high.
Was my fault, I thought the board's input clock was 33MHz, but it was actually 50.
That's assuming these can detect/debug baud rates.
Looks like something out of Nostromo equipment.
But why 5 cpus.... one for display/keyboard, one for interface/software but tye other 3? One for in, one for out, but then?
The "magic finger" strikes again! ;-)
How were you able to work out that C7 was the capacitor to replace on the PSU?
I'm guessing that either you were able to reverse engineer it or it's the same PSU design as the older devices. Thus did have a schematic available.
Its actually the easiest fix. Switching PSU that doesnt want to start or always restarts, 90% of the time its the small 10-100uF electrolytic on the primary side that also has a 50-500k resistor to charge it up from the high voltage bus. usually the cap goes bad, sometimes the resistor also goes up in value if it runs too hot.
Now that was easy.
Nice, I would chicken out and first dump all these ROMs before swapping them between ancient undocumented machines :)
Link to the ROM dumps are in the video description
Marc would it be possible to copy the newer firmware onto eeproms so both can do 80 columns higer res ?
I think my Kaypro x2 has a terminal emulation mode
many fond memories of this protocol analyzer some 35 years ago, when our network was RS232 and modems based (4800bps, bisync protocol)
What is the technical term for ( at start up if nothing happens use several fingers to activate many switches or buttons)
Is there a computer program that would allow something similar?
And can you control Ring port from c?
Built-in connector patch-panels!?! Is this real life, or this just fantasy
Nice!
Maybe you can copy the high res ROMs and transplant the copies intro the low res device?
Wouldn't that mean they have to battle the high resolution hardware fault? It's currently hidden because the low res firmware doesn't use it.
Great to see the success there :) I got IDACOM PT502 aka HP E3910C, WAN protocol tester. There is no documentation other than how to program it using FORTH. It has a MFM HDD in it (ST1100), and 7 CPUs, this time all Motorola: 5x MC68HC000 and two MC68EC030. No RS-232, the most mundane thing is RS-422, then it goes up to RS-449 and T1/E1. Software runs on those MC68000 CPUs, and it seems that 030s are used for handling E1/T1.
Unfortunately the disk failed in mine at some point. Do you think that Keysight would release the docs if I ask nicely?
I seem to recall that EIA-449 was actually the same signalling protocol as RS-232 but with more physical connector specs.
Do you think they will leave the 40 col one like that - lol -.
What is that floppy disk format in those drives - LS120?
Standard 3.5" DD floppy. Uses a proprietary HP format, not pc compatible.
👏👏👏👏👏
The decision to define the signals without defining the connector was a huge problem for human beings everywhere.
It wasn't all that bad. You just needed a breakout box and handful of adapters.
TNX AND 73
Duplicate the good ROMs and you will have two good machines that both do hi-res, right?
5 cpu and 7 fpga to analyse one serial line, I imagine each peripherals controller has his own cpu, keyboard, disk, serial line, led panel ? seems a little bit over engineered on the hardware side, and under engineered on the software side, don't you think that with a little bit of multi-programming (old term for multi threading), the M68K would suffice ?
It supports 2 serial lines running various higher level protocols. You are probably right that these days one high level CPU would receive two serial streams with two USARTs but when trying to do error detection on a serial stream it cannot offload the reception to a standard USART to determine the KIND of errors. The FPGAs would not have been doing much serious work except reducing the logic chip count to get everything fit onto the one main-board.
Also that is a CRT monitor so somewhere inside there there is analogue monitor electronics and the digital video controller IC or custom circuitry to handle the advanced character set switching.
It was a product of the times. Use $100 in parts that are known to work at the speeds required or use $10'000 in software that may not reach all the required speeds for all the advertised protocols. The earlier versions were mostly limited to 64k for the advanced synchronous protocols but the FPGA version may have supported a bit faster.
HELLO MARC. SHALL WE PLAY A GAME?
How about a nice game of chess?
These machines remind of the hundreds of machines AT's have in there shop to repair every kind of electronic Device used in aviation aircraft.. Thus why I choose not to be Electrician for NAVY went with Hydraulics... Awesome trouble shoot, about the same as removing flight control from aircraft avionics bay, Dropping on ground, (1m $) reinstall it fixed it..