SA Mixer repair - mixer differences

[orbanp1]

Hello Everyone,

The first mixer in my Polarad 632B Spectrum Analyzer is blown.
The Polarad 632B is a 2GHz SA, from around 1980.
The first LO to that mixer is ranging from 2GHz to 4GHz, the first IF frequency is 2GHz.
The mixer is "home built" by Polarad, the chances of finding a replacement is nil.
I would like to repair it.

My apologies if the following is a bit long winded, but please read on.

Enclosing a picture of the inside of the 632B mixer. It has a quad diode chip that would need to be replaced.
The original quad chip is an "Alpha D5847" quad array, unfortunately I could not find any data sheet on it.

Robin Whittle described the repair of such a first mixer of a Tek 7L14 SA, which is also a 2GHz SA.
The repair entailed the replacement of the single chip quad diode array:
https://www.firstpr.com.au/tequip/7L14/
For the Tek mixer repair Robin Whittle used the HP5082-2277 diode quad. That chip is still available from the Italian distributor, listed on that repair page.
However, because of the minimum order, I would have buy about ten of those chips.

A more economic alternative seems to be the Infineon BAT17-04WH chip which is a pair of detector diodes, good up to 6GHz, in a SOT323-3 package:
https://www.infineon.com/dgdl/Infine...6638963bf54e8f
That chip is inexpensive and is available at DigiKey.
The only thing that the chip contains only two diodes in series, but soldering two chips together at their two-leg side would make up a quad array, not much bigger than the original chip.
Any comment on making a quad array like that and using that diode quad as replacement?
If I have the skills to do the surgery and replace that chip successfully, that remains to be seen ;-)

The other question I have is about those Polarad mixers.
There are three such mixers on the RF-deck in the SA, they all look the same, I opened up two of them. Despite that the mixers look the same to me, and seemingly have the same internal structure, they have different part numbers. So what would be the reason for the different part numbers? Were they produced the same and then selected based on some criteria?

The SA itself is a triple conversion receiver, the second IF frequency is 65MHz, and the third is 10.7MHz. I enclose a block diagram of the SA.
The second mixer produces the second IF, the RF and LO frequencies are about 2GHz, with 65MHz difference.
The third similar looking mixer on the RF-deck is for phase locking the first LO when the sweep is applied at the third LO.
The phase lock mixer gets the 1st LO input and as the RF-input the output from a pullable comb generator in the same frequency range as the first LO. In this way the first LO can be phase locked to an X-tal oscillator at any frequency when that LO is in the CW mode. The first LO is in the CW mode when the frequency span of the SA is less than 100kHz/div, in this case the sweep is applied at the third LO.
The mixer for the third IF is on a card, not on the RF-deck, and it looks like an SBL-type DBM, though the part number does not look familiar.

As for the SA repair, I swapped the first mixer with the phase lock mixer and now the SA works.
Obviously I can not lock the first LO, but I do not recall that "Phase Lock" LED ever coming on on this unit, so that phase lock circuit probably has some other faults too.
But one step at a time.

Any comment, any help would be much appreciated!

Thanks, Peter

[Julesomega]

I'm very impressed with your hard work and results Peter. I'm just looking through my SOT23 Shottkys in case I can help: I have plenty of HSM-2822 (low flicker noise type) which is about the closest I can get to the BAT17-04W
I'll shout up if I can find any quad ring types

[John_BS]

This supplier has the HP quad in smaller quantities:


https://www.rf-microwave.com/en/hewl...ode/5082-2277/
Note that the 2276 should be almost identical.

John

[orbanp1]

Originally Posted by John_BS

This supplier has the HP quad in smaller quantities:


https://www.rf-microwave.com/en/hewl...ode/5082-2277/
Note that the 2276 should be almost identical.

John

Hi John,

That is the exact supplier that Tek SA repair page referred to.
They do have a minimum order requirement, that would be the order for about 10 chips.
Have not even looked at the shipping costs...

Thanks, Peter

[John_BS]

Peter; perhaps you could indicate what sort of budget you're willing to allocate to get this up & running.



Do you have a means to measure the LO power for the mixers? This might assist in indicating if the phase lock mixer is near enough the same to swap over. If they are the same, I'd be inclined to swap the phase-lock to the main input path, then either mend the blown one or buy a mini-circuits for the phase-lock position, which is less critical on linearity / frequency range etc. But before that you'd need to satify yourself that the overall performance with the phase-lock mixer substitution is close enought to the original spec.

John

[orbanp1]

Hi John,

Thanks for the reply!

Not sure about the budget, but getting that SA working is not that essential (but I would like to).
I do have another identical SA, I got two non-working units originally, and the other one is working now. It also had a myriad of problems, but luckily the RF parts were OK.
It has not been calibrated, but using a number of HP generators, and comparing the indication to a HP-141 SA and to a Cushman CE-15 signal analyzer, I am confident that the results are in the ballpark.

I did measure the LO of the first mixer at 2GHz, it is +9dBm.
I wanted to measure the drive level to the phase lock mixer too, I was also wondering if it has the same LO drive level.
I could not measure it, the output from that power splitter has a male SMA connector, I do not have the cable or adapter to connect to that. I need to get some of those!

The other minor issue I have is that I have the "B" version of the SA and the documentation I have is that of the "A" version.
In the "A" version all those similar mixers used single Shottky-diode quads.
But, among others, they have redesigned the RF-deck too in the "B" version...

Thanks, Peter

[orbanp1]

Hello Everyone,

Looks like the original thread is closed:
https://www.vintage-radio.net/forum/...t=mixer+repair
This is a new posting with the update on the mixer repair.

I have rebuilt the mixer, it seems to be working now.
The SA's calibration signal shows up with proper magnitude, with the harmonics also showing.
I have yet to make a full evaluation of the mixer performance, and compare the results with my other identical SA.

I needed to replace the diode quad in the original mixer.
The mixer at the LO and RF ports is running between 2GHz to 4GHz, at the IF-port from 100kHz to 2GHz.

For diodes I used Infineon BAT15 dual Shottky diodes. Those diodes are for mixers up to 12GHz.
The diodes in my quad are well matched the worst forward voltage difference at 1.3mA is 1mV.
The factory spec for a comparable HP quad diode claims better than 10mV match at 1mA.
I built up the quad on a fiberglass perf board that had plated hole vias. I used a 2x2 section, and removed any excess copper material from both sides that was not needed.
The parts are SOT323, real tiny, it is not fun to work with them.

One of the baluns is made exactly in the same way from thin twin-lines as in the Tek mixer that I referred to in my previous posting.
The small loop of the balun did not survive the disassembly, I made my own twin-line.
I used 30AWG wire warpping wire. Plugging in the geometry and relative dielectric numbers into a on-line twin-line calculator the characteristic impedance came to about 50 Ohm. I glued the wires to each other with rubber cement.

Here are some pictures of the mixer parts and an original, factory mixer.
(The original is a lot prettier!)

Peter

[Jon_G4MDC]

You can report your own thread using the exclamation mark and ask the mods to open it again. I expect they will concatenate threads and remove this post in good time.
(Nice work by the way!)

[Cobaltblue]

Thread Re-opened and new posts merged.

Cheers

Mike T

[orbanp1]

Happy New Year to Everyone!

I have made some measurements and further repairs on the SA, and it is nearly working now!

First of all, the mixer I rebuilt is not as good as an original!
I could adjust the vertical gain and offset in the SA to show the calibration signal at the proper level, and the harmonics are also showing up fully to the 2GHz limit.
However, when I adjust the gain and offset so the 10dB steps would match the screen 10dB markings, I needed to increase the gain so now the "grass", the noise level, is about halfway up the screen!
I have not made any further measurements on the rebuilt mixer, I just relegated it to the position of the phase lock mixer. It is working there satisfactorily.

I am just guessing that the conversion loss of the rebuilt mixer is not the same as for the original.
I could make such measurements, in-situ, I have tools to make the measurements up to about 1GHz. I might do that some later date.
Now I understand why do they select those mixers!

I also made some progress with the phase locking of the first LO.
I did measure and simulated every module in the phase locking circuit and could not find anything wrong.
As it turns out, the phase locking has been working all along, except for the "phase lock LED" on the front panel was not showing it!
The +12V is missing from the anode of that LED.
Unfortunately that is one change Polarad made from the "A" version to the "B", and I only have documentation on the "A" version.
In the "A" version the +12V comes from the board, in the "B" version, that I have, it comes from the mother board.
I need to track down that missing +12V!
Attaching a LED directly to the board shows the proper working of the phase locking feature.

I am not criticizing the Polarad design, but in my opinion the phase locking circuit design is marginal at best, as the simulations (and measurements) were showing it.
No wonder that is the feature they redesigned in the next "C" version of the SA.
Again, I have nothing but respect for the Polarad engineers, and it is easy to find faults in a circuit with 40 years of hindsight and with the computerized analysis tools available to us today!

In any event, I think I am making progress with the repair of this SA.

All the best for the New Year!
Regards, Peter

[Radio1950]

Good story.
The clarity of your technical writing is exceptional.

[orbanp1]

Hello Everyone,

I did find the missing +12V, it is coming from an extra deck of the rotary switch that selects the span width.
It provides supply to the LED only when the 1-st LO can be locked.
The switch did not provide reliable contact, exercising it solved the problem (- temporarily?).

There is one thing left, the digital memory "behind the screen" does not seem to be working.
Instead of a storage tube CRT this instrument has a 2k x 8-bit memory for storing the digitized data for the screen, in two pages.
The digital memory could store a single sweep, store the peak value of consecutive sweeps, and store/recall/erase a single page.
There is no microprocessor in this system, there is a board full with SSI and MSI logic circuits, and a control board with mixed analogue and digital circuits.
Only the "flow through" mode works in the SA where the digital memory is bypassed.
It is a complex circuit, it will take some time to understand its operation and testing it.

Regards, Peter

[orbanp1]

Hello Everyone,

Now the digital memory for the screen is working fine too on the SA.
The A/D conversion was not working, the controller chip for the successive approximation A/D converter, a Motorola MC14559 chip was at fault.
One of the control signal outputs was floating, it was open circuit.
The part took about three weeks to get here but now the SA is fully functional.

Thanks, Peter