Author Topic: EEVblog #529 - HP 35660A DSA Upgrade Investigation  (Read 21599 times)

0 Members and 1 Guest are viewing this topic.

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 37664
  • Country: au
    • EEVblog
EEVblog #529 - HP 35660A DSA Upgrade Investigation
« on: September 29, 2013, 09:41:23 pm »
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.
Datasheets:
http://www.farnell.com/datasheets/1465563.pdf
http://www.cotorelay.com/2900.php
http://www.ti.com/lit/ds/symlink/ne5534.pdf

« Last Edit: September 29, 2013, 10:37:25 pm by EEVblog »
 

Offline mimmus78

  • Supporter
  • ****
  • Posts: 676
  • Country: it
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #1 on: September 29, 2013, 10:12:04 pm »
Well, I have a similar project for my 2000 counts meter. Still not decided what opamp to use to change it to a 8.5 digits meter.  :)  :-DD

Sorry but i can't stop myself to post it :-)

Joking aside, it's a very good idea to improve specs of old equipment this way! Hope to see another follow up from this video.
« Last Edit: September 29, 2013, 10:25:39 pm by mimmus78 »
 

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #2 on: September 29, 2013, 10:30:10 pm »
Hi Dave,

Check the function of J10 and J20. It looks like they can be used to ground the input to the next stage. It would be interesting to measure the spectral noise density with the jumpers in the grounded position. It will give you an idea about the noise contribution of the front end.

Jay_Diddy_B
 

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 37664
  • Country: au
    • EEVblog
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #3 on: September 29, 2013, 10:38:18 pm »
Check the function of J10 and J20. It looks like they can be used to ground the input to the next stage. It would be interesting to measure the spectral noise density with the jumpers in the grounded position. It will give you an idea about the noise contribution of the front end.

Ah, didn't see those, interesting, thanks.
J200 even bypasses the anti-alias filters and other stuff, right at the track/hold input. Very nice. Someone made this thing really easy to measure stage by stage.
« Last Edit: September 29, 2013, 10:48:16 pm by EEVblog »
 

Offline c4757p

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #4 on: September 29, 2013, 10:41:48 pm »
Completely OT... but I have to say, I noticed in the schematic there that they've bolded the main signal path. I've not seen (or noticed, more likely) that before and I really like it :-+
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 2731
  • Country: ca
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #5 on: September 30, 2013, 12:22:50 am »
Hi,
I just did a quick check of my HP 35665A DSA, this is the slightly newer version of the 35660A.

With the input terminated in 50 Ohms, I measured a PSD of -153 dBVrms/rtHz

which corresponds to 22 nVrms/rtHz.

Conspiracy? two similar instruments half a world apart, one answer  :-+

Since full scale is -51dBV

This means that -153dBVrms  is -102 dB relative to fullscale.

Theoretical SNR = 6.02N + 1.76 dB where N=number of bits.

So 102dB SNR corresponds to 16.6 Bits

This with an averaging of 100 samples.

So I don't know how much improvement you are going to get from the front end. I am interested to watch  :-+


Jay_Diddy_B



 

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 37664
  • Country: au
    • EEVblog
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #6 on: September 30, 2013, 08:26:23 am »
I tried the two jumpers which bypasses the front end. No difference at all. same 22nVrms/RtHz result.
I also discovered that my analog board is very significantly different to the schematic and overlay in the service manual. The left input side of the board is the same, but the right and side (by the analog ground plane spit) is vastly different. Doesn't appear to have the same transitor based sample and hold circuit, and no J200 jumper to bypass all the anti-alias filter etc.
 

Offline varno

  • Newbie
  • Posts: 9
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #7 on: September 30, 2013, 10:16:29 am »
The best Low-Noise op-amp that I know of on the market is the LME-49990 at 0.88nv/sqrt(Hz), that amp has noise that is lower than most discreet JFETs, though if you want a low noise FET, one of the best that you can get is the 2SK170, they are hard to get, but if you want a few and are willing to find a good matched pair I'll send you a bunch for you to have at it.
« Last Edit: September 30, 2013, 10:18:51 am by varno »
 

Offline sync

  • Frequent Contributor
  • **
  • Posts: 799
  • Country: de
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #8 on: September 30, 2013, 10:57:31 am »
Replacing a monolithic dual JFET with two discrete ones sounds like a bad idea.
 

Offline varno

  • Newbie
  • Posts: 9
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #9 on: September 30, 2013, 11:10:33 am »
IF you match them for Ids, and then glue them together with thermal epoxy, it works well. so long as the transistors are from the same batch.
 

Offline JackOfVA

  • Supporter
  • ****
  • Posts: 350
  • Country: us
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #10 on: September 30, 2013, 11:30:32 am »
My HP 3562A DSA, one generation earlier than Dave's machine, shows 34.6 nV/sqrt(Hz) at 160 Hz, with 50 ohm termination and 128 sweep average.  (Averages must be in powers of 2 with the 3562A).

Specs are:
20 Hz - 1 KHz <-134 dBV/sqrt(Hz)  corresponds to 200nV/sqrt(Hz)
> 1 KHz  < 144 dBV/sqrt(Hz) corresponds to 63 nV/sqrt(Hz)



 

Offline sync

  • Frequent Contributor
  • **
  • Posts: 799
  • Country: de
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #11 on: September 30, 2013, 11:32:39 am »
IF you match them for Ids, and then glue them together with thermal epoxy, it works well. so long as the transistors are from the same batch.
I doubt that. This is not audio stuff. The Analyser goes down to 244uHz.
 

Offline varno

  • Newbie
  • Posts: 9
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #12 on: September 30, 2013, 12:18:18 pm »
You would be surprised as to how badly monolithic dual-jfets can be matched.

If he wants to go and get a single matched part the 2SK389, is there as an option, or the obvious course of action would be to use the op-amp on it's own. it is actually very good,
 

Offline Salas

  • Frequent Contributor
  • **
  • Posts: 292
  • Country: gr
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #13 on: September 30, 2013, 06:53:30 pm »
I tried the two jumpers which bypasses the front end. No difference at all. same 22nVrms/RtHz result.
I also discovered that my analog board is very significantly different to the schematic and overlay in the service manual. The left input side of the board is the same, but the right and side (by the analog ground plane spit) is vastly different. Doesn't appear to have the same transitor based sample and hold circuit, and no J200 jumper to bypass all the anti-alias filter etc.

From the schematic's areas you passed over in the video it seems there is no gain where you focus, it looks like just a combo buffer. The gain stage is somewhere next.
 

Offline Salas

  • Frequent Contributor
  • **
  • Posts: 292
  • Country: gr
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #14 on: September 30, 2013, 09:10:41 pm »
Could be the 2N5546 JFETs.
 

Offline EE59

  • Newbie
  • Posts: 1
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #15 on: October 01, 2013, 12:43:21 am »
Check the function of J10 and J20. It looks like they can be used to ground the input to the next stage. It would be interesting to measure the spectral noise density with the jumpers in the grounded position. It will give you an idea about the noise contribution of the front end.

Ah, didn't see those, interesting, thanks.
J200 even bypasses the anti-alias filters and other stuff, right at the track/hold input. Very nice. Someone made this thing really easy to measure stage by stage.

I don't think J10 and J20 ground the input to the next stage.  They enable the bootstrap circuit to dynamically follow the output and adjust power to the FETs, allowing them to handle higher voltage swings.  With them grounded you could overdrive the FETs.  They shouldn't change the spectral noise density. 

Most of the noise is from the series input resistors R9, R10, and R20.

That bootstrap circuit would make a good fundamentals Friday video.
« Last Edit: October 01, 2013, 10:05:19 am by EE59 »
 

Offline casinada

  • Frequent Contributor
  • **
  • Posts: 599
  • Country: us
 

Offline tecman

  • Frequent Contributor
  • **
  • Posts: 444
  • Country: us
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #17 on: October 02, 2013, 02:52:55 pm »
I would try AD797 op amps.  Internally compensated, more than enough bandwidth and 0.9 nV/rt-hz.  Should drop in as a replacement, just remove the existing comp cap.

paul
 

Offline hobbs

  • Contributor
  • Posts: 15
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #18 on: October 04, 2013, 04:03:45 am »
I have the 35665A.  There's an outfit (glkinst.com) that has reverse-engineered the options system in that one, and offers an All Options EEPROM to turn them all on.  Works great, and it was pretty cheap at $65 including a new CMOS backup battery.


Cheers

Phil Hobbs
 

Offline vk2hmc

  • Contributor
  • Posts: 24
  • Country: 00
    • VK2HMC
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #19 on: October 08, 2013, 05:36:31 am »
Dave,

here is a hint!

If the JEDEC part number is not listed in the products part list description, use a HP Xref to convert HP part numbers to JEDEC (www.jedec.org) part numbers.

For your Part, 26-0715 (18 is dropped).
Therefore, the HP part number for the Op-Amp is 1826-0715.

Using one of the HP Xref's, this equates to a NE5534!

The Original HP cross reference was posted in HP bench briefs regularly so there are quite a few different scanned copies around.
I also triple check part numbers before ordering/replacement, just to be sure!


This is my main Xref - http://www.vk2hmc.net/blog/?wpfb_dl=7 (!warning 21Mb)
Be aware here are a number of different cross reference lists present in the file.
If the part is not in the 1st list, be persistent and keep searching.

Also, I find that this searchable excel spread sheet is pretty quick, accurate and fairly complete - http://www.vk2hmc.net/blog/?wpfb_dl=6

Actually, once you get used to the HP part system, it is quite easy to both remember and find the JEDEC part number if you have to.



--marki
« Last Edit: October 08, 2013, 01:59:38 pm by vk2hmc »
 

Offline JoeO

  • Frequent Contributor
  • **
  • Posts: 527
  • Country: us
  • I admit to being deplorable
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #20 on: October 08, 2013, 01:29:14 pm »
Marki:

Thanks for the lists.  They are great.

You forgot to mention that the first list IS searchable in a PDF viewer.

I did look for the 1826-0715 in the PDF file.  It is listed in there 6 times as a NE5534AN. 

Joe
The day Al Gore was born there were 7,000 polar bears on Earth.
Today, only 26,000 remain.
 

Offline vk2hmc

  • Contributor
  • Posts: 24
  • Country: 00
    • VK2HMC
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #21 on: October 08, 2013, 02:01:21 pm »
triple  :palm:!

I got it into my head the part is a 1826-0711 for some reason even though I _wrote_ 1826-0715 in the post - go figure...

 

Offline innkeeper

  • Supporter
  • ****
  • Posts: 615
  • Country: us
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #22 on: August 17, 2017, 01:05:28 am »
I found a cross reference for the 55-0460 / 1855-0460  it showed a 2n5545

anyone know of a less noisy replacement for it?

this fet's noise is typically rated at:  20 nv/(sqrt)hz
2n5545 data sheet http://www.vishay.com/docs/70253/70253.pdf

I'll post the link to the cross reference when I find it again, but it confirms someone suspicion from earlier

This topic caught my eye as I just scored an HP 3562A  :-+ should be arriving shortly.
the service manual does have a bom and part numbers, and seems to share a very similar front end to the HP 35660A and uses the same part numbers.

I'm more then willing to give upgrading the opamps and fets it a try once I do the initial checkout of it.

Anyone think there would be any reason not to socket the opamps with machined sockets like mill-max?





« Last Edit: August 19, 2017, 04:26:46 pm by innkeeper »
Hobbyist and a retired engineer and possibly a test equipment addict, though, searching for the equipment to test for that.
 

Offline innkeeper

  • Supporter
  • ****
  • Posts: 615
  • Country: us
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #23 on: August 19, 2017, 08:14:26 pm »
Looking at the schematic, it appears to me the 4 2n5545's on the front end of the high and low buffer side would introduce the most noise earliest in the input at 22nV/rtHz.

Someone correct me if i am wrong, but it would seem if one can't make any noise floor improvements by upgrading those, that likely any otter op amp upgrades later in the circuit will likely not net much.

i do notice they drive an offset adjustment. using (A33/A35) R212 and 1412 in the HP hp 3562A. I couldn't find the procedure to adjust those, does anyone know how thatch done?
Also any commendations on good replacements for the 2n5545?

Hobbyist and a retired engineer and possibly a test equipment addict, though, searching for the equipment to test for that.
 

Offline chuckb

  • Frequent Contributor
  • **
  • Posts: 339
  • Country: us
Re: EEVblog #529 - HP 35660A DSA Upgrade Investigation
« Reply #24 on: August 21, 2018, 02:18:36 pm »
Linear Integrated Systems has a very low noise dual JFET the LSK489. It seems to have about 4 times less noise than the 2N5545. The capacitance seems similar also.

Then the noise limit would be the NE5534 op amps. An LT1028A would be a good replacement for that. the Compensation may need to change to keep it stable. Or use the LT1128A.

That's just in the front end buffer, I don't have a schematic to identify other downstream opamps that may introduce noise.

I have an HP35665A that I want to reduce the noise on. Let me know if any of this helps!

Mill-Max sockets should be fine at these frequencies.

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf