HP 5087A Distribution Amplifier has a mystery board
Hello and welcome back. If you follow the channel, you know that we have been working on getting enough HP 5061A atomic clocks working, to attempt a redo of the famous Hafele and Keating direct measurement of relativistic time dilation, which was done by flying 4 of these very clocks around the world in 1971. But before we get there, we need to build a measurement setup to track tiny time drifts between our atomic clocks. Although there is specialized test equipment that can do that,
I don't have it, so we are going to build our own, you guessed it, out of a stack of vintage HP test equipment. Some of which we still need to restore before we can use it. So welcome to a mini restoration within a larger restoration, working on the HP 5087A distribution amplifiers. As the name implies, this instrument is used to distribute the output of one reference frequency to many instruments, something we'll need so we can synchronize all of our measurement instruments to one reference atomic clock.
[Marc] So, the 5087 is one of those forever instruments. It's first mentioned in the 1972 catalog with tentative specifications. But lo and behold, I found it in this catalog, although by then they have become too cheap to include a picture. Because,
this is the 1999 catalog! I don't know if they produced it afterwards, into the 2000s, because I don't have any catalog that's newer than this. But, made for 30 years or something like that! And, on the front of the instrument, there's not much. There is two lights, to indicate if it's on. There's no on/off switch, as soon as you power it up, it's on. And then, you have a little meter that enables you to monitor your three inputs, and your 12 outputs, and your supply. And, at the back, is the actually the business end of the instrument. You have the three inputs, that you distribute to 12 outputs. And how the
three inputs are distributed to the 12 outputs, is actually sort of unique to each instrument. It can be configured. So they are not two the same. And, usually, in the front, there's a little display. I remade this one to indicate what's in it. This is future Marc, so we'll see it in the rest of the video, this one has some interesting stuff inside. Well, thanks future Marc, I just realized I had not described the instrument front and back. So now, back to the regular Marc.
[Marc] And those are immensely configurable things, which means they're not two the same. It's for distributing your clock signals, from a reference source, to plenty of user units. But it's meant to distribute three channels.
I was a bit mystified by this three input scheme. But then I looked over to our 5061A atomic clock, and it started to make sense. The 5061A has three outputs: 5 MHz, 1 MHz, and 100 kHz. So the 5087 was made to match that.
By the way, all three frequencies are a bit annoying. The 100 kHz is also a historical hold over, for driving the 115 BR divider clock, that we restored previously. The 1 MHz is too slow, and the 5 MHz is not fast enough for modern instruments.
As we'll see, the HP 5087A can help with that, since it can do frequency division and multiplication internally. There are three buses inside, and you have to connect them to output amplifiers. One of the most simple config is 5 MHz in, and then you have four of the output 5 MHz amplifiers. Then 1 MHz for the 1 MHz outputs, and 100 kHz for the 100 kHz output. But you can get crazy, because you can distribute one input to everything. So that's another popular option. And another one is that they have doublers and dividers. And then,
you can have the 5 MHz come in, then go from bus A to bus B through a divider. So, that creates the 1 MHz, and then goes from bus B to bus C to another divider that creates the 100 KHz. So you can have the same as this, but using only one input. What I want is actually the contrary, I want a doubler.
Both the quartz and the cesium have this annoying thing, that they just give you 5 MHz, and I need 10. And there is a doubler option. I think there's one of them in one of those boxes. So, I have to first figure out what's in each of those boxes, because they are not two the same, of course. So, two at the top are Marcel's, two at the bottom are mine. So Marcel has one unit which is the classic 5, 1, 100 KHz, and another one that was half 5 MHz, half 10 MHz. So, this one has, actually a few cards. So, four 5 MHz board, and...
But it has three inputs. One is a 91034. What is that card? How many screws does one need? 1, 2, 3, 4, 5! A50! So that's the one that's bizarre, an A50. And that must be a more recent thing, because I haven't seen it mentioned anywhere. Okay. Okay, clean, clean! How does that work? The whole module goes up? Yep, the whole module goes up. And that's the unidentified card.
So here I am looking at one of the three input modules, the ones that sits on input channel C. Usually they are either input amplifiers, dividers, or multipliers. But this one is unusual. We are having a fancy module here, with modern electronicals in it. What the heck? HP 5087, so it's made for this instrument. Dash 91034. For reference, that would be a regular amplifier module. This is input module that was in channel B, a simple preamp, the most common type.
That's when they used gold liberally! So, it's a simple video amplifier with a gain adjustment. So over there was the input amplifier that I just put back in place, and that's the output amplifier. Also tons of gold. So people just go after that and destroy them to get the gold. This is one of the output amps. They work at only one frequency, there are output amps for 100 kHz,
1 MHz, 5 MHz or 10 MHz. A chassis has a max of 3 input modules and 12 output modules. Except ours, that has two regular input modules and a 3rd one that we don't know what it is. Gold diggers, which is kind of sad, but... Okay, so let me figure out what this is.
So, I spent way too much time trying to figure out what this is. And I think I have an idea what this is. The amplifier here is the same as this, or very similar.
You can see the big inductor, and if you follow the schematics, it's about it. They just shrinkified that with tantalum capacitors. So, it is an input amplifier. But, in front of it, the circuit which I thought would be dividers, they are not! They are op-amps, inverter gates, and an analog switch - a four-way analog switch. And what I think this is, after thinking of it, this is a failover card. That switches, on the output, either A or B. One application could be, if you have a clock that needs to be driven by something, and it's mission critical, right, you're at NIST or something.
Well, if one clock goes down, then you switch over to the next clock drive signal, and you won't lose a beat. And I think that's what this is. I couldn't find any mention of it in any HP catalog, but there's at least one other one for sale on the web.
I might complete the reverse engineering, just to be sure. A little longer than a few minutes later... I reverse engineered it, and this thing is this noodle over here, which simplifies into this.
And this is actually exactly what I thought: a failover card. That's the same amplifier as normal. They put a little amplifier for the meter, that doesn't exist in the original card. Fine. And then, this amplifier has now two inputs: one is from input B, and the other one is from input A. And they're controlled by analog switches, which are in this chip over here. There's four of them, but they only use two.
And the way it's controlled is that. If you follow input A, it goes into a complicated bit of circuitry. This is just a level detector, a rectifier, to figure out the level of signal in A.
And then it goes in here, that's the little adjustment here, and you can decide below which level it will fail over. And so, there's a little amplifier, and then it goes into a comparator. And if it falls below the level, then switch A is open and the other switch is closed. So you switch over from A, which is normally the one that's on, to B. So input A is the normal and input B is the fail over. And C is not used as an input. It's used as an output for an alarm signal.
So, when it has failed over, it will just copy the signal out. And then you can light a bulb, or do something with it. So that you know that it has failed over, get an alarm . It would be kind of cool to put the cesium on that one, and the ovenized quartz on this one. And just demonstrate that you can zap the cesium, and the clock will continue on the quartz, and demonstrate a fail over.
Interesting card! It is weird that it's not documented anywhere. All in all, it was time well spent to reverse engineer that card, because I discovered that it was configured in a little bit of an inconsistent way. This is a regular amplifier, and it was in slot A, and configured to take its input from A.
If you can look at them, they have little zero ohm jumpers. This one in slot B amplifying input B, which is correct. And then the third channel, channel C, is this weird fail over thing that monitors A, B and then outputs C. Those basically were not really used at all, because this card in slot C, it will output on C. C is correct, C is correct, C is correct, this is incorrect. This should be a C.
Or you could choose to have it monitor just A, the primary, or the backup. You don't want to mix the three signals at the same time. Here we go. And a one. Oh, this one fell out, okay. Okay, I hooked it up, and it seems like everything is working as it should.
So if I plug it in, there's no on/off, it goes on right away. And you could check the power supply, it's good. Now I have hooked two inputs. I have my primary input, it's coming from the scope at 5 MHz exactly. And then I have a secondary input, coming from this generator. I have
detuned it a bit at 4.9 or 4.8 MHz, so we can see the difference. It should now be in the primary mode. So it should take its input from A, spit it on to C, all those cards are amplified to C. And I see the signal.
And now, I'm trying to do a fail over. I'm going to turn off this generator from the scope. And the other one, my little HP, is going to take over. And it did! So, now, you can tell it's at a slightly different frequency. I did that on purpose. Well you can't see it, but it says 4.8 MHz.
And then, the input C, which is indeed used as an output, went low. And now if I turn the generator back on, this signal went back up, and I'm back on the primary generator. So that that all seems to be fine! Interesting a little gizmo. Well, I discovered it's not all perfect. All the cards don't have a good level. Channel 3 is down. Is it because it is out of tune? Oh yeah,
it's completely out of tune! That's why. And channel 4 is okay. Okay, so fortunately I have the right extender. I think the only thing I have to do, is get this peaked. Oh my, I was so far away!
We have returned THE signal to an appropriate value. Yeah, it's super high. And we even wrote it all up, so the next HP enthusiast doesn't have to spend half a day like us to figure out what this is all about. So this unit is done.
And while I was at it, I even made a little one of the stickers that you're supposed to have - the other one has it - that describes the configuration. So, I think I got all the brownie points on that one, it's done. So, let's get to the second one, which is actually more interesting to me. So, this one appears to be a standard configuration. Actually,
not quite. It's 1 MHz, 5 MHz and 10 MHz. Which is good, because I'm interested in a 10 MHz. There is a doubler card that goes with this instrument. Unfortunately, I don't think this has one, but I have a solution for that. And... So, supposedly that's the regular broadband amplifier. And it says...
Yeah, it's an 0464. And I can see the jumper is on A on this one. The jumper is on B on this one, and the jumper is on C on this one. So, they are properly configured. Probably those three cards, those four cards, are configured correctly for picking the A bus. Man, if I can take it out. This is indeed a 1 MHz, and it's picking up the output of the A card, bus A. I'm going to check them all, that they are properly configured, but I think, since this is a factory configuration, it should be fine.
Yes, this is on A. So, four cards properly configured at 1 MHz. This one, I suppose is, a five. And it's configured to take output from B. Yes, this is B. Yes, this is five.
By the way, to change one from a 5, from a 10, it's really easy. They're the same cards. So let's look at the schematics for the output amp cards, kindly provided by HP. Here is the 100 kHz card, then the 1 MHz card, scaled to sort of fit on top of each other. Look closely as I flip between the two. 100 K again, 1 MHz again. And here is the 5 MHz card, and the 10 MHz card. Notice anything similar?
Yes! They are essentially the same thing, except for the capacitors here, which are two 5600 pf in parallel on the 100k card, and just a single 800 pF on the 1 MHz card. They simply change the resonance frequency of the tank circuits formed with T1 and T2, from a 100kHz to 1 MHz. Similarly, the 5 MHz has 160 pF caps, and the 10 MHz has 22 pF caps. In addition, on the 10 MhHz, R7 is 82 Ohms instead of 100. Maybe for stability margin, although I don't
think it really matters. So, if your eBay unit does not have the cards you hope for, no problem! Just change the caps for the right frequency and retune it. And the input amplifiers are broadband and work across the whole range, no need to adjust them, there is just one type. So, they are the same card. The only thing that's different about it, is the capacitor here. So,
you can very easily change one to the other, by changing to the right capacitance. So, if you have too many of one kind, not enough of the other, very easy to change them to the right frequency. And now I'm done. There should be... yeah, this is wired to C. And this should be my precious 10 MHz. Yep, that is all fine. All right, so, now the question is: I only have 5 MHz coming out of my clocks, and I want to be able to have 10. And there used to be a card that would have a doubler on it,
but I couldn't find any in the wild or on eBay. However, there's a clever chap that has remade his own version, and sells them on eBay for not not too expensive a price. It's SCSkits on eBay, and it is a 5 MHz to 10 MHz. And the only
thing you have to be careful of, you have to jumper them properly. And I jumpered it for C over here, right there. And so, I can put it in my C slot here, so it will get the input from C. And that should give me 10 MHz, if I am correct. And I've set my wave gen at 5 MHz already. So, with a little bit of luck... Pam! This is too easy!
So, here, we got it! And it says 10 MHz. So it just works! That's neat. Yeah, plenty of signal. And I can check if it's tuned properly.
Huh, okay, let me see if I can get the tuning a little bit better here. It's working! I'm getting better at adjusting this. You need to add a 50 Ohms termination if your scope doesn't have it.
And 1V RMS, which is 2.8 Volts peak to peak, which is what I have here. The first thing you want to do, is adjust the output of the preamplifier. Here it's way too high, it should be 0.3, not 1. So it actually takes very little. So, 0.3, there it is. Then you want to look at that board, five. You want this to be around 3V. Oops, I'm getting confused...
Yeah, it was way too high. So 3 volts RMS over here. So 0.3 on the input, 3 on this one. And now I have to change my scale here on the FFT. 2 MHz, okay. First let, me tune it. They were all a little bit out of tune.
Ah well, this one was pretty much right on. So for once, one that's good! And then let's see if I can minimize the harmonics here. Yeah, right here. It's this one over here, the second harmonic. Right there. So now this card is good. It's on three volt over here, it has minimized distortion, it's at the tuning peak.
Then you have to do that for every card, so it takes a while. A few minutes later... Here we go, we have a fully calibrated HP 5087A! All right, I even updated the label to reflect the configuration, which actually was not factory, it was not even what was on the label.
So, let's go and use it to make fancy experiments. Well, that was a bit of a detour, but I wanted to cover this instrument for the record, as it is a bit of a classic amongst time nuts. We still managed to reverse engineer a very cool and undocumented failover card, and learned how to configure and tune the thing. Foremost, we added a doubler card, so we can distribute an atomic-derived 10 MHz reference to our other instruments. Time to make some fancy metrology measurements! See you in the next episode!
2025-03-14