A teardown of the front end PCB in the HP 35660A Dynamic Signal Analyser.
Dave does some preliminary investigation to see what opamps and FET's are unsed in the analog front end to see if they can be upgraded for better performance.
This is more for Dave's record than interesting info.
Previous video: http://www.youtube.com/watch?v=Y0jkPLuFdnM
Forum: http://www.eevblog.com/forum/blog/eevblog-529-hp-35660a-dsa-upgrade-investigation/'>http://www.eevblog.com/forum/blog/eevblog-529-hp-35660a-dsa-upgrade-investigation/
Datasheets:
http://www.farnell.com/datasheets/1465563.pdf
http://www.cotorelay.com/2900.php
http://www.ti.com/lit/ds/symlink/ne5534.pdf
EEVblog Main Web Site:
http://www.eevblog.com
EEVblog Amazon Store:
http://astore.amazon.com/eevblogstore-20
Donations:
http://www.eevblog.com/donations/
Projects:
http://www.eevblog.com/projects/
Electronics Info Wiki:
http://www.eevblog.com/wiki/

Hi in my previous tutorial video on Opamp voltage noise where I used my HP 356a Dynamic signal analyzer here to uh measure the noise floor of some Opam s the power spectral density really of some opamps and I mentioned that uh I might investigate actually upgrading this thing. uh because it's noise floor is not great I mean we're talking about um, you know in the order of 22 uh Vols per root Hertz uh noise flow of this thing at 1 khz there, so not very spectacular. So I uh talked about the possibility of um opening this thing up. have a look inside and see what Op amps and components are used in the front end there and see if I can maybe replace them with some uh, modern, more modern ones really, you know, designed for Ultra ultra low noise performance and see if I could, maybe, uh, do a drop in replacement for some of those opamps? perhaps uh to increase the performance of this uh DSA it's certainly possible.

and I do have the service manual for this thing, but unfortunately it doesn't actually tell you on it. It's got full schematic and PCB overlay, but it doesn't have a bomb reference and the schematic doesn't actually tell you which particular opamp is used on the front end which is really annoying. So I'm going to have to actually open this thing up. uh take out the main board which I didn't do in a previous video where I uh actually repaired this thing didn't take out that analog board so I thought I'd take it out, have a look see which Uh chips used in it and well go from there.

So this video is more of just like a uh documenting uh reference for myself really to uh uh, document the uh noise floor um as it stands here on this thing uh, the original condition and then uh, look at the parts used and then if we do the upgrade then we have uh to see if we can improve the performance of it. Now for the reference measurements of the noise floor in this thing: I've got Channel One range fixed to Uh minus 51 DB volts RMS that is the lowest range it's 4 M volts uh Peak voltage range on the input so that's all fixed. uh I'm measuring the Uh Power Spectra uh spectral density of this thing and um, as the previous video I actually made I noticed I made the mistake of um the vertical units here here instead of volts RMS per Roots Herz I actually used volts per Roots volts per root Herz and that gave me a figure before of around 31.4 whereas I should be using volts RMS per root Rootz So that's what I'll use as the reference here and at 1 khz there my marker frequency. this is over a frequency span of uh 1.6 khz there.

um and uh basically if you enter 1 khz as the span then it actually says oh, I can't do that I'll default to 1.6 and we're getting a reference value there with a 50 ohm terminated input on channel one of 22.2 Nan nanov volts RMS per root Hertz And by the way, that was for a floating input and with Uh DC coupling. Now I'm actually measuring uh channel two here and you can't just select channel two. you got to select dual mode. But uh, there we go.
We're getting little bit higher, but it hasn't done the 100 averages yet. It's a little bit noisier there on Channel 2 by the looks of it. 24.5 6 Nan volts RMS per Roots Herz for reference on channel 2 over 1.6 khz span and that's the figure with a 12.8 khz span and 100 khz span with 10 khz marker. And by the way, these are all performed with Uh flat toop window in so I can repeat that.

Uh, the 100 khz one with a handin window. There we go for reference and that's the 100 htz response once again. uh, flat top window and that's the 10 Hertz response after only 23 averages because it does take 32 seconds. uh per record length.

So it takes quite some time to get to 100 here. that'll probably do and we got 163 Nano volt per route Hertz at 1 Hertz there. All right, let's see if we can take uh, the Channel One board out here. So I'm going to have to unscrew these rails here, take off the ribbon cables and hopefully it'll lift out, but there's got to be a coax going through to the BNC on front panel.

So I hope I don't have to uh, take off the front panel and then do the B SE housing and all that sort of jazz. we'll see Actually I don't even think I have to take that board out because this um shielded top just uh lifted uh, straight off and you can see the shielded hand in there. But I believe one of those two opamps down there is the front end opamp. I can see, you probably can't see it down there.

Oh yeah, there we go. A couple of metal can transistors down in there I believe they're the fet buffer front end. um, but I believe one of the first. Well, the first Op amp in the chain is one of those puppies.

and if we have a look at the Uh schematic here of the front end, please excuse the lack of screen capture here. Here's the BNC on the input over here and uh, then we have all of our read relays to do the various Um AC DC coupling shorted out input. uh, test signals, things like that uh, input attenuation path uh input 50 ohm uh termination down here and stuff like that and then it goes into check this out. Here we go.

There's our looks like they've got a Fet, a matched Fet um input buffer there with an unknown opamp doesn't actually have the number on it I Kind of assumed it might have maybe been a generic um Ne5 34, but uh I just looked at the number in there and that doesn't look to be the case because there is no bomb here. so that's U1 and U2 it's a differential thing. So they got a low and a high uh fet buffer there. So instead of using a Fet input up amp, they've used a fet um, uh, they've used a matched uh matched fet pair on the input and effectively turned it into a fet input.

uh Op amp because fed input opamps generally have higher uh input noise uh, voltage density than um, your than your bipolar opamp. So um, you know, really, that's that's maybe why they've gone for that fet input. I mean I Don't know what the state of-the-art in uh, you know, fed input Op amps was back in when this was designed back in the late Um 80s and probably even before that carry on from uh, previous uh DSA designs. I'm not sure, um, because you know this isn't one of their oldest Dsas, not by a long uh, stretch.
But anyway, um, yeah, so that opamp is, probably you know as a first guess without knowing what this transistor is, because as I said, they don't have a bomb for this thing. All they got a component overlay and nothing's marked on there. so uh, we'd have to look at the metal can there to figure out the transistor, but probably as a first guess. I would say that that Iant maybe dominates um that in terms of the buffer there and then it goes to the next uh sheet over here.

But if we go to our component overlay, here's our component overlay and this is what we looked at on the board there. Here we go. there's our two opamps on the top U1 and U2 So they're definitely our two Op amps and with inside that front end shielded can there that is just um, all of your real lays K There they are. and there's Q1 and Q2 So there's your two matched uh transistor pairs there.

so they're inside the can so that would be annoying. We have to take the can off probably to see the part number on those suckers. but let's go back to those uh dip chips there U1 and U2 and they are not uh, ne5 34s. they are a signetics.

Um, what you know, almost looks like a maybe HP part number on that sucker. We may be able to track that um I think as a first guess 27715 but it's got Fgk 2563 on it that is not familiar at all. You can see the date code there first week 1990 and they've used those also in a couple of locations 2563 there and there as well I Don't know where they are in the circuit, but uh yeah, there's certainly quite a few of them. Tada We have the board out.

It wasn't easy to get out, but uh, it did come out yet. No coax used as a beautiful little uh, right angle BNC there to connect down to a mating BNC right down the bottom which then has the coax which goes to the front. Just beautiful design. Absolutely stunning.

I Love it! So uh, the high-res photos of this by the way will be up on my flicker account for those interested. Well actually, it wasn't hard to find out. uh what this mystery labeled Ch is? Yes, I was right. it is just a signetics N534.

Um, classic audio opamp. Um, but it's got some weird ass uh, you know custom HP part number on it and it's used all throughout this thing. There's like a dozen of them on this board. um reason I Know that is cuz that's U4 and we can see U4 on the schematic and it is one of the rare chips in there that we is actually labeled N534 so it's got to be it if we just have a quick uh look at the input front end here.

um nice little Cotto relays I'll have to check out the uh part number on those, but look at that beasty that's actually the uh 50 ohm input Terminator go figure and those uh codo Reed Reays of course Cotto One of the uh premium brands in the business 2900 series Re relays exactly what we want. There are hermetically sealed of course, epoxy coated shells provides magnetic shielding, optional electrostatic Shield to reduce capacity of coupling, and optional coaxial cable uh Shield as well. so you know really purpose design for one of these uh, front ends. It really is quite a nicely laid out board.
As you'd expect, look at the big ground split right down there. Beautiful! I Love it how you can see through these uh old style boards. They're just superb quality. I Love them.

and yeah bloody. HP part numbers on everything. Real pain in the ass. look at that.

and I count 17 of of those signetics Ne5 34s on there. There's a whole stack of them over in this section here. They've almost use those exclusively for the opamps and there is under the metal shield and you'll note that uh yeah, they've removed the ground plane from under there as well. Uh, and there's our you know, it's basically all of the input uh switching.

It comes in down the bottom here of course and then uh so a signal comes in here from the front panel, got our 50 Ohm Terminator and the relays to switch that on and the relays to ground the input and stuff like that. and then we've basically got our uh, front end stuff in here. This is all our high impedance fet stuff and then these are our two up up the top here. They're our first Op amps combined with that fet buffer.

But it's interesting to note that they actually have um, put the Fet inputs under the shielder can and haven't put the opamps under there. You'll notice a couple of guard traces going from from just directly on the metal can there around, probably around the back of the Um input pins to the opamp and the both the fets on the input. They've got 55046 labeled on them and a National Semiconductor and another mark on the other side. and there it is.

National Semiconductor D 8952 So I'm not sure what that part number that actually translates to yet. haven't been able to find anything on that but these Um input up amps definitely um, N534. So unless they're really specially selected from Signetics I Doubt it. Um, because this is not a particularly uh, low noise unit by any uh, stretch as we've seen, then um, yeah there at 1 khz I think about 3 nanovolts per root.

Hertz can certainly get a lot better than that. uh these days in turn, like a Ad797 or something like that. but uh, not really directly pin compatible replacement because there is compensation on those. a compensation used on those amps Um, which may not be compatible with any replacement ship.

You just have to be very careful. So I don't know I'm going to have to uh mle this one over. haven't got enough time to uh mull it over today, but uh, a shame they're not in sockets. Of course it would have been really easy just to you know.
Suck It And See Really? Um, well. I could certainly do that, suck off the solder and see no pun intended. Um, it might be worth a uh try. So you know, really, the input noise is probably going to dominate in the front end here.

combination of the Uh fets plus the opab and it's going to be the input Uh current, uh noise as well as the input voltage, noise, combination of both and the whole thing there. Um, so you know it might be worthwhile. Uh, you know, trying to upgrade this section, see if we can get better performance. So anyway, if anyone has any uh uh thoughts on that, please please leave it in the comments.

Catch you next time.

Avatar photo

By YTB

28 thoughts on “Eevblog #529 – hp 35660a dsa upgrade investigation”
  1. Avataaar/Circle Created with python_avatars RB says:

    Found it. Great video!

  2. Avataaar/Circle Created with python_avatars Navarro Eletrônica says:

    Is this just a nice fast FFT?

  3. Avataaar/Circle Created with python_avatars Alex MacKellar says:

    Analyze the frequency of that loogie bouncing around in your voice.

  4. Avataaar/Circle Created with python_avatars Милан Павловић says:

    How you achieve full view on CRT?

  5. Avataaar/Circle Created with python_avatars Julian Weinert says:

    Hey, how did you actually fix the dead edges of the screen? Didn't see that in the repair…

  6. Avataaar/Circle Created with python_avatars Me Mer says:

    Torrent 😉

  7. Avataaar/Circle Created with python_avatars KX36 says:

    It's black. At and around 10:47 you can see black text around the board, and the black line down the middle in question branches off to the right clearly at 11:05 crosing several top layer traces. So the line must be silkscreen dividing areas of the board for some reason.

  8. Avataaar/Circle Created with python_avatars Shit, I Missed. says:

    he's already ranted about people like you complaining about his voice. while his rants are entertaining, I do not wish to see dave's brain release the magic smoke

  9. Avataaar/Circle Created with python_avatars Andor Márton Horváth says:

    Dave, the problem is, more modern op-amps than the AD797 usually only have SO-8 packaging instead of DIP-8. If you can upgrade the PCB using SO-8 op-amps, you could try the LME49990 from TI, it has better than great specs, and it doesn't need compensation as far as I remember.

  10. Avataaar/Circle Created with python_avatars sparkyuiop says:

    "Oscability"! Nice word, but is it a word? It should be.

  11. Avataaar/Circle Created with python_avatars John Richards says:

    O yes your right, That takes us back to mikes question.

  12. Avataaar/Circle Created with python_avatars Tom Bruhns says:

    Yes, you're right: the 4 FETs do contribute about 20nV/rtHz at 1kHz, and that would be improved by using better FETs. Nonetheless, you'll quickly get to a point where noise is dominated by the output stage of the input buffer op amps, because the input buffer stage is run at unity gain. And of course you'll also be limited by the noise in those 5k resistors.

  13. Avataaar/Circle Created with python_avatars 11zekim says:

    Siliconix datasheet shows 10 nV/rHz typical @ 1 kHz. Two FETs per amp, two amps per channel (high and reference) gives 20 nV/rHz added in quadrature. Add 7 nV/rHz * sqrt(2) for the two 5k series protection resistors, I get 22 nV/rHz RTI.

    And it's all unity gain; no allowance made for the diff amp, AA filter or ADC.

  14. Avataaar/Circle Created with python_avatars Tom Bruhns says:

    Look again. It's black, and hard to see against the dark green part of the board where there's a ground layer; much easier to see in the shielded area where Dave has the shield off.

  15. Avataaar/Circle Created with python_avatars Tom Bruhns says:

    I should add: the 2N5545 is indeed not particularly low noise, but it's lower noise than Dave's seeing. You'd gain a _little_ by replacing it with a more modern, lower-noise JFET pair, but you'd be disappointed that you didn't get down into the low single-digit nV/rtHz spec'd for your part. You probably WOULD improve the noise at very low frequencies significantly; the 2N5545 has a pretty high 1/f noise corner.

  16. Avataaar/Circle Created with python_avatars Tom Bruhns says:

    Be really careful trying to second-guess what the designers had in mind! There's a clear reason why the noise is as high as it is, and it's not the FETs. It has to do with a design choice to achieve good performance in another area. It would be easy to miss that point and mess up the performance.

  17. Avataaar/Circle Created with python_avatars Peter-Paul Vervoort says:

    Maybe the previous video: watch?v=Y0jkPLuFdnM is more to your liking. And if not maybe Analog electronics is not your thing (true for many I guess 😉 ). But give it a try, once you understand a new world opens up to you 🙂

  18. Avataaar/Circle Created with python_avatars Peter-Paul Vervoort says:

    You're correct if the FET stage would have sufficient gain. But this circuit looks like a voltage buffer to me so gain = 1 so the opamp will most likely be the largest noise contributer.

  19. Avataaar/Circle Created with python_avatars Tom Bruhns says:

    Dave, you might note that the power to the dual FET is bootstrapped to follow the input voltage. There are some implications there to why you see so much noise. It's not a BNC: it's an SMB! Typical HP part numbers: 1826-xxxx are analog ICs, 1820-xxxx are digital. So the 26-0715 really is 1826-0715. Sometimes you can get lucky with a google search or similar, as you can with this one. If you get really lucky, you score an HP cross-reference booklet. (1853=PNP, 1854=NPN, 1855=FET)

  20. Avataaar/Circle Created with python_avatars Tom Bruhns says:

    They're JFETs; not sure why the strange symbol with an arrow into a floating gate! Someone really should have put some voltage gain in that buffer; the noise is the result of output noise in the op amp, not the noise of the FET or input op amp noise.

  21. Avataaar/Circle Created with python_avatars Tom Bruhns says:

    Seems like an easier way to get lower noise (and what I did, in fact) is put a low-noise ext preamp in front of it. You can even use the HP35660/35665/35670 to measure the gain of the preamp as a function of frequency, store that in a register, and use the math functions to more directly display calibrated readings. My application was for a low-impedance source; I used an AD797, 2 9V batteries for pwr, and well shielded, to keep the preamp noise very low. For AC cplg: 10uF of polyprop caps.

  22. Avataaar/Circle Created with python_avatars SwitchingPower says:

    Dave the HP 1855-0460 matched FET on the input is a 2N5545 "Monolithic N-Channel JFET Duals"

  23. Avataaar/Circle Created with python_avatars John Richards says:

    I cant see any silkscreen.

  24. Avataaar/Circle Created with python_avatars Nikola Jambrović says:

    Hi
    Have you found anything about those input FET-s?

  25. Avataaar/Circle Created with python_avatars Mike Whitenton says:

    Dude, you nailed it! Take a look at 4:51. There is a shield that runs right down that line. They didn't want the shield to wear through to the traces so they ran all of them on the back of the board. I didn't even notice it during the disassembly. Dave was talking about the BNCs and I wasn't paying attention to the shield. Good call on that one.

  26. Avataaar/Circle Created with python_avatars nickoe says:

    It could possibly be because some mechanical wall or so goes there.

  27. Avataaar/Circle Created with python_avatars ReabowRotors says:

    you should put in a socket if you do try an upgrade

  28. Avataaar/Circle Created with python_avatars KX36 says:

    AD8675 looks like a nice part, which I had previously overlooked. Quite similar to the OPA209 I've used before. Cheers.

Leave a Reply

Your email address will not be published. Required fields are marked *