Hewlett Packard HP461A Wide Band Amplifier

[Peter_O]

Not shure if this one should be here or in 'Beginners'.
It's grown up lab equipment, but so far my knowledge might be kind of beginners level. 

2023 this little guy needed to be rescued for 30 Euro.



The 461 provides a gain of 20 dB or 40 dB with a frequency response of +/- 1 dB from 1 kHz to 150 MHz and a max output of 0.5 V RMS into 50 Ohm.
https://www.keysight.com/us/en/assets/9018-02904/user-manuals/9018-02904.pdf

Obviously there have been some repairs in this one already.

 



What bothers me are these points:

1)
The PSU voltage is in spec, but the ripple is 8 mV RMS instead of < 1 mV spec.
Would you change the caps?

2)
The bode plot shows quite OK behavior below 10 MHz but goes wild above that.

 

Following the service manual I tried to adjust the inductors, but there was no change in the frequency response at all, so there might be broken coils.

3)
The transformer is running quite hot with 60 °C as well as some resistors. Is this normal for this kind of device or do I need to search for a fault?

 

Has anybody of you been an one of these points before?
Hints etc. appreciated!

[Stray Electron]

  Nice save!  What's the date code on that unit?

   Those small silver capacitors shown in picture #4 (and some in Picture #2) are wet slug tantalum caps and are VERY expensive and very high quality and rarely fail. I would not touch them unless you can prove that they're bad.  But I would check those large silver electrolytic caps.

  I've never worked on one of those but IMO the transformer should not be getting hot. 

  Looking at the second thermal image it looks like that capacitor that's partially hiding under the black heat sink is getting very hot. I sort of doubt that it's bad but I think something is overloading it.  (Possibly that amplifier is oscillating and is dumping high frequency into the capacitor).

[Peter_O]

  Nice save!  What's the date code on that unit?

Thx for your ideas! 
It's a 0946... .

   Those small silver capacitors shown in picture #4 (and some in Picture #2) are wet slug tantalum caps and are VERY expensive and very high quality and rarely fail. I would not touch them unless you can prove that they're bad.  But I would check those large silver electrolytic caps.

Will check the PSU caps.

  I've never worked on one of those but IMO the transformer should not be getting hot. 

  Looking at the second thermal image it looks like that capacitor that's partially hiding under the black heat sink is getting very hot. I sort of doubt that it's bad but I think something is overloading it.  (Possibly that amplifier is oscillating and is dumping high frequency into the capacitor).

Will check the temperature of the cap and try to do some more measurements.

[coppercone2]

mine had a cracked attenuator resistor that was driving me nuts. Its slightly out of spec now by 1dB but I can live with that.

[Stray Electron]

  Nice save!  What's the date code on that unit?

Thx for your ideas! 
It's a 0946... .

  Wahoo! Built in 1969.  Nominally the "46" week of the year but HP actually marked a date about 8 weeks (IIRC) AFTER the actual production date to allow time for shipping, etc.  So your's was probably being built at about the same time that Apollo 11 was landing on the moon!

[factory]

  Nice save!  What's the date code on that unit?

Thx for your ideas! 
It's a 0946... .

  Wahoo! Built in 1969.  Nominally the "46" week of the year but HP actually marked a date about 8 weeks (IIRC) AFTER the actual production date to allow time for shipping, etc.  So your's was probably being built at about the same time that Apollo 11 was landing on the moon!

It wasn't built in 1969, the serial prefix code is the date code of the last major design revision.

Why else would it have four components showing 1978 date codes and another showing 1977? There are probably many more not visible from the pictures.

There are other clues such as the printed HP logo, post 1970 changes, IEC socket & white plastic power indicator light and later cover colors.

David

[Swainster]

I've got one of these, sort of - sadly a previous owner had plundered the output transistor and attenuator for parts. I've got it partly working by trial and error substituting various RF transistors for the missing HP unobtanium part. Other than the missing bits, the main issue was that a lot of the carbon composite resistors had drifted out of spec. As this operates up to the VHF band, I wasn't convinced that metal film resistors would be suitable replacement for carbon composite so I put SMD resistors on some sliced up copper clad as the closest substitute that I could come up with from my 'stash'. Lastly, I can say that if the RF tuning slugs are stuck, don't force them as they turn back to dust as the slightest provocation. Still a (stalled) work in progress, as the flatness is out of spec (crumbled slugs not helping here) and the attenuator is still missing some mechanical bits.

BTW, member Tony_G did a youtube video on his example.

Regarding your bode plot results, could the issue be a dodgy BNC connector?

[Peter_O]

Let me add a clear schematics pdf.

I've checked two of  the bigger PSU caps C1, ... C5:

They should be 65 µF and are
82 µF / 0,5 Ohm and
79 µF / 0,6 Ohm
at 100 Hz.

So I think, all five should be fine, or do you think the ESR is to high?

C7, one of the smaller PSU ones:

It should be 20 µF and is
22 µF / 0,11 Ohm.

-----

And I took another look at the hot 220 Oh resistors R52 and R53 near the output emitter follower Q9. They are still hot at 80°C.

snip: Looking at the second thermal image it looks like that capacitor that's partially hiding under the black heat sink is getting very hot. I sort of doubt that it's bad but I think something is overloading it.  (Possibly that amplifier is oscillating and is dumping high frequency into the capacitor).

This is C30, which is the final output coupling capacitor. Having a closer look, it definitely stays cold.

[MathWizard]

So if this was built with discrete 1970's part's, besides the expensive caps someone mentioned, how good of a version can be built with today's jellybean parts? What about for say $50, case/hardware not included ?

I want to try it with my cheap parts, I haven't looked at the schematic yet, maybe it's some JFET buffer input like the JFET buffer probe I made. My cheap JFET's might be ok for 50-100MHz

[factory]

   Those small silver capacitors shown in picture #4 (and some in Picture #2) are wet slug tantalum caps and are VERY expensive and very high quality and rarely fail. I would not touch them unless you can prove that they're bad.

The Kemet T110 are hermetically sealed solid tants, can't tell what the others are from the picture. Have you checked for any that might be shorted? it does happen, just not as often as the infamous bead type.

Does the transformer still overheat if the secondary windings are disconnected and did you measure power consumption? Also check the fuse is the correct value for the voltage setting.

David

[David Hess]

So if this was built with discrete 1970's part's, besides the expensive caps someone mentioned, how good of a version can be built with today's jellybean parts? What about for say $50, case/hardware not included ?

A modern inexpensive implementation would use several operational amplifiers in series and have a frequency range from DC to several MHz, similar to the example below.

[Someone]

So if this was built with discrete 1970's part's, besides the expensive caps someone mentioned, how good of a version can be built with today's jellybean parts? What about for say $50, case/hardware not included ?

I want to try it with my cheap parts, I haven't looked at the schematic yet, maybe it's some JFET buffer input like the JFET buffer probe I made. My cheap JFET's might be ok for 50-100MHz

Its 50 ohm input, so just find a MMIC amplifier with suitable flatness down to DC. $1-2 ?

[Peter_O]

   Those small silver capacitors shown in picture #4 (and some in Picture #2) are wet slug tantalum caps and are VERY expensive and very high quality and rarely fail. I would not touch them unless you can prove that they're bad.

The Kemet T110 are hermetically sealed solid tants, can't tell what the others are from the picture. Have you checked for any that might be shorted? it does happen, just not as often as the infamous bead type.

Does the transformer still overheat if the secondary windings are disconnected and did you measure power consumption? Also check the fuse is the correct value for the voltage setting.

David

Yes. I woll disconnect Amp from PSU and check separately in the next step.
Thx for the hint to check the cap(s) for shorts, David!
Will do and come back here.

[coppercone2]

did you do a continuity check on the coils?

if so, desolder them and LCR test? its a bit of a wild goose chase. i bet you can replace the inductors.

[Peter_O]

The main fuse is in spec: 250mA.

Disconnected secondary side of transformer.
Resistance is
primary:
1-2 240 Ohm
2-3 0 Ohm
3-4 270 Ohm
secondary:
5-6 31 Ohm
6-7 34 Ohm

Voltage selector is OK:
1-4 is 230V
Secondary Voltage is
5-6: 44 V RMS
6-7: 44 V RMS

Disconnected Trafo stays cold.
Scope with differential probe shows same voltages ...

 

As R52 and R53 were getting hot, I followed David's (factory) advice and checked C29 (orange).
It's 2.2µF bang on and not shorted.



With connected transformer and seperated +15V and -15V lines between PSU and amp,
I put a 2k resistor as a load in between.
Output is +14.6 and -16.2V.
Trafo stays cold.
But R2 and R3 (yellow) are getting hot quickly (100°C).

Next step is to better understand this PSU. 

And to put an lab PSU at the amp and see, what happens.

[coppercone2]

i mean the tuner coils

[Peter_O]

i mean the tuner coils

Sure.
Tuning the amp will be the next step.
After having solved the problem of running 'too hot'.

[David Hess]

Another thing to think about when looking at this design is that transistors were the most expensive part, so all effort was made to minimize how many were used.  A modern discrete implementation could use cascode stages, doubling the number of transistors, but producing a flat response without the coil peaking.

[coppercone2]

i mean the tuner coils

Sure.
Tuning the amp will be the next step.
After having solved the problem of running 'too hot'.

I just meant to check if there is broken wire on the coils maybe it snapped. see if there is zero ohm continuity across it

[coppercone2]

Another thing to think about when looking at this design is that transistors were the most expensive part, so all effort was made to minimize how many were used.  A modern discrete implementation could use cascode stages, doubling the number of transistors, but producing a flat response without the coil peaking.

i noticed smart people are united against coils lol. like seriously, they wanna get rid of coils. the silicon design guys get angry at inductors

i wonder if they will have advanced meta material or refined structure inductors one day that was used by anti transistor zealots in RF designs. or some kind of 'super inductor'

but this amp weirded me out. I used it with an antenna on a tecsun radio. it seemed to help the signal quality (but this was one with a damaged attenuator). I heard some guy from the west coast talking on HAM and it felt very eerie, like he was present at my house. it made me wonder about psionics. He also happened to be talking about the nature of perception lol. Then I took it apart to fix it and recently finished fixing it. I am just waiting to find a transcript some where that the conversation that I heard took place in 1965 and that the man has been dead for 50 years. Or that he was aware his transmission was being used a test signal thousands of miles away by a certain  and that my bank account details go. is this box amplifying more then photons?? turns out someone lifted the schematic from a classified CIA fun file titled 'stargate'. I never had the feeling someone might guess my bank account details because of an amplifier but I swear I turned it off because it felt like some how the dude was in my house lol

[David Hess]

I noticed smart people are united against coils lol. like seriously, they wanna get rid of coils. the silicon design guys get angry at inductors

i wonder if they will have advanced meta material or refined structure inductors one day that was used by anti transistor zealots in RF designs. or some kind of 'super inductor'

Until we get superconductors, inductors just do not perform well compared to capacitors.

More advanced designs of that era used inductors, but not for extending the bandwidth of each stage in such a simple way.  Tektronix had the secret of bridged t-coils for many years.

[coppercone2]

I think there might still be something with like powder technologies related to inductor improvements. There is a SURPRISING amount of science behind powders that most EE are not aware of, save for maybe super capacitor / battery process people, an a surprising amount of liberty taken in device production that ignores MANY details/possible measurements of even unprocessed feed stock used to make some kind of structure, let alone the assembly process. Not sure what effect any of it would have but it would be expensive and time consuming to determine / control some known factors. and I bet there is even more behind magnetic properties of powders.

[factory]

This probably won't help, but here are some pictures of my 462A, the circuit diagram in a previous post shows the differences, it's a bit older, there are some Ge transistors in the earlier ones, the coils are well sealed to prevent adjustment.

The manual only mentions Q9 having a heatsink fitted, mine is exactly as described, the 461A seems to have a load of extra heatsinks for unknown reasons.

I don't like the use of a 250mA fuse for both 115V & 230V operation, especially with that tiny transformer.

David

[MathWizard]

Well I'm going to make something like this, I'm just trying to get a model working to spec's in LTSpice with my transistors. I'm trying to find a higher quality image of the BJT list. I can tweak the biasing a bit if needed. Or try to, I've never tried to calculate anything for such a baising, I wonder what those DC and AC eqn's will be.

I'll power it from a bench PSU. Or maybe try making a low noise, low power split rail PSU just for little test devices/probes.

I ordered some 0805 LCR SMD sample-books, with lots of small nH inductors and C0G/NP0 caps. I made a JFET buffer probe that works ok. Now I need to miniaturize it and fit it in some metal tube and get some coax.

[MathWizard]

Can anyone make out the BJT's in the part list ? Q4 is 2N3478, I could try S9018 for most of these. But for the PNP, I don't think I have anything fast like a 9018.

I would like to try adding JFET's to the input, like a buffer probe, but for the class AB amp, I still need faster PNP's. I just made a big Digikey order, but I forget fast PNP's