A friend of mine has a Jerrold VHF/UHF sweep generator which I believe was made in the late 1940s. The thing is built like a tank; the tuning mechanism in the thing is so overbuilt that it looks like a rack-and-pinion steering box.
Awesome teardown. From the comments I see that some people still seem to believe that 3 GHz is in any way high or impossible to deal with on a hobby scale. That is absolutely incorrect. With nowadays tools it's nothing. Amateur radio operators build circuits all the way to 47 GHz and higher with amateur equipment at home. One just has to know how.
I think that's the OCXO you've got, those CTS oscillators are common on eBay in similar packages, I've got one or two slightly different ones, both with ovens. Awesome teardown. I took apart my 8657A (all-analogue signal path, I think) and was shocked by how much *stuff* it takes just to generate a sine wave. This one goes from +10 to -143dBm. Staggering indeed. If you consider the whole 153dB range, we're past Jupiter. For 1AU I get 112dB, btw.
Electrostatic RF MEMS relays are still an active research area. They're mostly used in satellites and military applications. They also require 50-100V to actuate. My guess would be that gold module contains a very well characterized and shielded semiconductor attenuator.
You're right. While it is totally possible to just reference two voltages to each other using the standard dB-formula, it is almost always the squares of the amplitudes that get referenced to each other (that's where the 20 comes from) exactly for the reason to give the same dB values for voltage and power.
I bet this was used in an automated test rack either for high volume production testing or extended reliability testing. If you notice, not only was it on for 44,000 hours but the attenuator also has over 14 million cycles on it.
Mike - notice that the log screen shows "attenuator cycles" so I suspect that there ARE indeed relays in the attenuator. They may be very tiny and not audible outside their enclosure.
Or it could be that the count is a hang-on from an earlier version - there is a service about retrofitting the attenuator with the "electronic" version.
Very cool teardown! This may be a silly question, but how would one reliably get a 3 GHz signal from the generator to whatever is being tested when even a couple millimeters of conductor has enough inductance to mess with it?
Coax cable works fine, albeit with some losses. Don't ask me about the physics - to do with transmission lines & such. Everything being low impedance also helps a lot.
Yes, the formula confused me, too. It is correct, except for the "1dB". The number would depend on what is substituted for p1 and p2. It should rather be L(dB) = 10 x log (p2/p1) or L(dBm) = 10 x log (p2/1mW).
The general formula for dB is L(dB) = 10 x log (p2/p1), where p1/2 can be any physical quantity: power, voltage, pressure, aso. So dB really is just a ratio. But the important thing to understand is that dBm is something slightly different. It means your p1 is 1mW (and therefore p2 has to be a power, too). So dBm is just a shorthand for writing "dB referenced to 1mW". Look up decibel on Wikipedia. There's a list of common dBs.
Good job with the dB analogy! These distributed filters and other circuitry really are 21st. century magic, beautiful and creeeeeeepy at the same time. Agilent: The voodoo whoodoo what you don't daredoo people....
Thanks for the explanation about the db-scale. I always wondered about that, how a purely relative value could clearly specify an output voltage. So, is this "0db=1mW" some kind of standard value among high frequency gear?
Those tracks may meander, but they certainly are not lazy. I think the moral of the story is that there are always a few random passives on the bottom side!
Is it just me, or is this like most arcane and mystic skill of human kind to design and build bleeding edge RF designs? I comprehend basics and even some details of almost any hard engineering discipline, be it rockets, computer chips, bridge, cryptography, chemistry, quantum mechanics, but magic voodoo always eludes me. I guess, the things about RF is that you cannot easily decompose it into blocks, and the every little details like placement and angel of computes matter, making it almost an art.
and speaking of absorption, they have these shielded cells on the board presumably to keep EMI from one section going into the next but they don't seem to have any RF absorbing material in there so you have this really strong echo chamber for every section which I can't see as all that great considering anechoic chamber designs. Or is my thinking on that wrong? If not then maybe RF isn't as hard as they make it look.
Hi Mike! Nice teardown as always. I read in some of your comments that the seller from whom you bought the signal generator also has some nice PSU. I am interested in buying a PSU, could you please tell me who the vendor is? Thanks!
at 3.2GHz, aren't some of the chambers big enough for that? and even if not a resonant chamber, does the radiation energy disappear? would encasing in absorbing material or skin tight conducting shielding be better?
I wonder if you could just do RF circuits really tightly with little regard for EMI and then just coat it with a thin insulator material and then encase it in a thick graphite resin to absorb whatever stray stuff my arise. and then metal foil outside that. Even though RF requires good science, once designed it could conceivably be dirt cheap to produce. I assume none of the components are made from Britney Spears virgin tears which would be hard to come by.
To more familiar with software or digital stuff (not this silly RF voodoo), 140db difference is equivalent to 46-47 bits of resolution. This is ENORMOUS. You hardly get 24 bit in most advanced DAC. Usually 16. And usually 8-12 in very high speed DACs. 46 is like 1 million time better than 24 bit most advanced audio DAC or DMM ADC.
Probably the little uncalibration that give you not correct at the end of the video, it cause when you dissasambled the instruments and the somthing is not in the same as before.
1. Stop using metal tools to point at electronics 2. Place the camera on a tripod, I got sea sick trying to watch this 3. Stop mumbling, I barely understood half of what you mumbled.
A friend of mine has a Jerrold VHF/UHF sweep generator which I believe was made in the late 1940s. The thing is built like a tank; the tuning mechanism in the thing is so overbuilt that it looks like a rack-and-pinion steering box.
Awesome teardown. From the comments I see that some people still seem to believe that 3 GHz is in any way high or impossible to deal with on a hobby scale. That is absolutely incorrect. With nowadays tools it's nothing. Amateur radio operators build circuits all the way to 47 GHz and higher with amateur equipment at home. One just has to know how.
I think that's the OCXO you've got, those CTS oscillators are common on eBay in similar packages, I've got one or two slightly different ones, both with ovens.
Awesome teardown. I took apart my 8657A (all-analogue signal path, I think) and was shocked by how much *stuff* it takes just to generate a sine wave. This one goes from +10 to -143dBm. Staggering indeed. If you consider the whole 153dB range, we're past Jupiter.
For 1AU I get 112dB, btw.
love the analogy at the end. That's what I enjoy most about your videos, you always throw that something extra in.
Electrostatic RF MEMS relays are still an active research area. They're mostly used in satellites and military applications. They also require 50-100V to actuate. My guess would be that gold module contains a very well characterized and shielded semiconductor attenuator.
You're right. While it is totally possible to just reference two voltages to each other using the standard dB-formula, it is almost always the squares of the amplitudes that get referenced to each other (that's where the 20 comes from) exactly for the reason to give the same dB values for voltage and power.
GBP750 - still a few on Ebay UK for that price. Same seller also has some nice PSUs
I bet this was used in an automated test rack either for high volume production testing or extended reliability testing. If you notice, not only was it on for 44,000 hours but the attenuator also has over 14 million cycles on it.
Mike - notice that the log screen shows "attenuator cycles" so I suspect that there ARE indeed relays in the attenuator. They may be very tiny and not audible outside their enclosure.
Or it could be that the count is a hang-on from an earlier version - there is a service about retrofitting the attenuator with the "electronic" version.
Very cool teardown!
This may be a silly question, but how would one reliably get a 3 GHz signal from the generator to whatever is being tested when even a couple millimeters of conductor has enough inductance to mess with it?
Coax cable works fine, albeit with some losses. Don't ask me about the physics - to do with transmission lines & such. Everything being low impedance also helps a lot.
Yes, the formula confused me, too. It is correct, except for the "1dB". The number would depend on what is substituted for p1 and p2. It should rather be L(dB) = 10 x log (p2/p1) or L(dBm) = 10 x log (p2/1mW).
Yeah - I only noticed this after doing the vid. the fact they count cycles suggests maybe it's something mechanical - MEMS relays perhaps?
The general formula for dB is L(dB) = 10 x log (p2/p1), where p1/2 can be any physical quantity: power, voltage, pressure, aso. So dB really is just a ratio. But the important thing to understand is that dBm is something slightly different. It means your p1 is 1mW (and therefore p2 has to be a power, too). So dBm is just a shorthand for writing "dB referenced to 1mW".
Look up decibel on Wikipedia. There's a list of common dBs.
Good job with the dB analogy! These distributed filters and other circuitry really are 21st. century magic, beautiful and creeeeeeepy at the same time. Agilent: The voodoo whoodoo what you don't daredoo people....
Thanks for the explanation about the db-scale. I always wondered about that, how a purely relative value could clearly specify an output voltage. So, is this "0db=1mW" some kind of standard value among high frequency gear?
I'm guessing that sharp bends are inductive and radio transmissive
Those tracks may meander, but they certainly are not lazy. I think the moral of the story is that there are always a few random passives on the bottom side!
A wonderful find, that piece of gear. Thanks for the tear down...
Is it just me, or is this like most arcane and mystic skill of human kind to design and build bleeding edge RF designs? I comprehend basics and even some details of almost any hard engineering discipline, be it rockets, computer chips, bridge, cryptography, chemistry, quantum mechanics, but magic voodoo always eludes me. I guess, the things about RF is that you cannot easily decompose it into blocks, and the every little details like placement and angel of computes matter, making it almost an art.
24:36 Hmm. Don't you mean 0 dB =, or to put the 10x inside the log?
and speaking of absorption, they have these shielded cells on the board presumably to keep EMI from one section going into the next but they don't seem to have any RF absorbing material in there so you have this really strong echo chamber for every section which I can't see as all that great considering anechoic chamber designs. Or is my thinking on that wrong?
If not then maybe RF isn't as hard as they make it look.
Trust me, Aglient/HP knew what they were doing back then and they know what they are doing now.
Electronic engineering.... a science
High frequency....an art form!
Nice piece of gear! How much did you pay for it?
Nice bit of kit!!
Hi Mike! Nice teardown as always. I read in some of your comments that the seller from whom you bought the signal generator also has some nice PSU. I am interested in buying a PSU, could you please tell me who the vendor is? Thanks!
at 3.2GHz, aren't some of the chambers big enough for that?
and even if not a resonant chamber, does the radiation energy disappear?
would encasing in absorbing material or skin tight conducting shielding be better?
Mike, did you mean to say 0.2 PPM @ 21:40?
Your agilent can transmit to the sun?
What the heck has been eating the front of your workbench?
Drill mostly.
Thanks for making this Videos, this is great.
Awesome video!
Do you pick your stuff up from auctions? How much did you pay for it?
I wonder if you could just do RF circuits really tightly with little regard for EMI and then just coat it with a thin insulator material and then encase it in a thick graphite resin to absorb whatever stray stuff my arise. and then metal foil outside that.
Even though RF requires good science, once designed it could conceivably be dirt cheap to produce. I assume none of the components are made from Britney Spears virgin tears which would be hard to come by.
awesome! really enjoyed that one!
Knife?
0db = 1 aU?
Nice vid!
Just a bit tired
search 6632B on ebay UK
To more familiar with software or digital stuff (not this silly RF voodoo), 140db difference is equivalent to 46-47 bits of resolution. This is ENORMOUS. You hardly get 24 bit in most advanced DAC. Usually 16. And usually 8-12 in very high speed DACs. 46 is like 1 million time better than 24 bit most advanced audio DAC or DMM ADC.
Please don´t show us parts with matal screwdriver, I am getting heart attacks.
Your videos are amazing but I think it's time get/use some decent lighting.
Nice ending! :D
Probably the little uncalibration that give you not correct at the end of the video, it cause when you dissasambled the instruments and the somthing is not in the same as before.
Thanks!
eeprom for calibration contants.
Coax.
Woouh, dude. that is seriously wrong! Power is defined as 10 * log and voltage/current as 20 * log.
Built like a brick shit-house!
And another electronics pr0n video.
1. Stop using metal tools to point at electronics
2. Place the camera on a tripod, I got sea sick trying to watch this
3. Stop mumbling, I barely understood half of what you mumbled.
Shut the fuck up bitch.
Are you sick? Your voice is different, its weak...