HP 8551B Spectrum Analyser

[Diabolical Artificer]

Bought a non working HP8551B a while back, popped it on the bench, the 851B display section seems ok, but the RF section has a duff LV tfmr. Have a feeling that it may have been powered up too long with bad power regulator boards, bad smoothing cans. There was a 3mm bolt in place of the AC line in fuse; not a good sign.

I briefly powered it up with a LL/variac to take voltage readings, but T2 the LV tfmr started smoking, power off sharpish. Un-soldered all the wires and found low DC res readings on several windings, so T2 kaput.

T2 the so called LV tfmr has 166 0 166v, 165 0 165v, 212 0 212 83v and 6.3v secondary's with a 0-110 (120) 0 - 110 (120)v primary. Joy of joys, it's potted, so no easy rewind.

The first problem is without de potting, finding the current ratings for the secondarys is going to be tricky. The LV section supplies + 150v @ 128mA, -200v @ 47mA, + 300 @ 31mA. The 6.3v I can work out by adding the valve's V1 & V2 current requirements.

I planned to try and get this fixed so all the PSU's are ok, and pass it on to someone more experienced, but I'm wondering if it's worth it, as there are probably a lot of other issue's. however, I'm not one for scrapping nice old bit's of gear like this without a fight.

I've attached T2 and it's regulation circuits, if someone could give it a once over and give me a rough estimate of the AC current of it's secondary's, that'd be a big help.

Cheers, Andy.

[orbanp1]

Hi Andy,

These are series regulators, so the DC current would be the same as the RMS current from the secondary transformer windings.
As the transformer has center tapped secondaries with two way rectification (except for that 6.3V winding) the RMS current would be half of that on either side of the center tapped winding.
Use some safety factor to size up these currents a bit.
You could also use the total power consumption of the instrument to double check the calculations.

Any chance of finding a same size transformer that is not potted, and rewinding that?
Would be a shame to scrap that analyzer.

Regards, Peter

[Diabolical Artificer]

Thanks Peter, I'm getting my head round this a bit more after reading the manual. With the PSU reg circuit being a bit busy, it threw me.

I presumed the two 166 0 166 (336v) windings are two 150v supplies stacked, as mentioned int manual the 212 etc being the - 200v. It's weird one is given as 166 and tother 165, maybe a typo.

I looked in my tfmr stash and didn't have one of the same size or any laminations that size.

I managed to get the tfmr apart, but not in a nice way. Popped it in the oven for an hour to try and soften the potting compound, nada, so had to use brute force (cold chisel and lump hammer). So today I should be able to measure the old one and re use the core. It's way over wound and had a copper band/screen on it, despite being in a thick tin. Mind the tin was steel so wouldn't stop stray magnetics. I should be able to make a pretty good replacement, but will use ali instead of amour plate.

Had a quick squizz at the circuit after the tfmr, found several bad SC caps de- coupling those diodes, that would certainly give the tfmr a hard time.

A.

[orbanp1]

Hi Andy,

Just looked at the schematics again, the voltage that I called 6.3V is indeed 83V! (The one "stacked at the top of" the 2 x 212V winding.)
So take that into account! (I guess I need new glasses ;-)

Just another idea.
As you mentioned, the circuits are quite complex, probably because of the lack of high voltage semiconductors at the time of the design.
These days there are lots of semiconductors that handle those voltages as well as high voltage three point adjustable regulators.
Possibly you could re-engineer the PS with only a few modern components.
I also do understand that is not what you have the problem with, but it possibly could make the design and winding of a replacement transformer a bit easier.

Regards, Peter

[Diabolical Artificer]

That's an idea Peter, mosfets would regulate the piddling needs of the LV section easily. I did wonder why they used half wave rectified centre tapped secs, bridge rectified would make the windings a lot easier. However I think there's some protection circuitry in there too, to cut the anode supply of the BWO for instance. I'm wading through the manual, it's huge and mostly incomprehensible to a relative novice, and gleaning the odd nuggets of relevant info.

Dismantled the tfmr, it fought me every step of the way, it being impregnated with varnish. First i had to cut through the winding's and bobbin with an angle grinder to get the first lam off; not what I wanted at all. Then every lam had to be cut free with an old knife, took me all morning. They usually just fall to bits once you have the first 20 or so lams off with a gentle clonk from a hammer.

The laminations are very thin and not made from the usual steel I think. I'm a bit concerned I may have altered the magnetic property's by warming it up. The lams arn't rigid like spring steel which is what the tfmr lams I've dealt with in the past have been like.

In the process of making a bobbin, had to make it an 1/8" smaller as I destroyed about four lams unfortunatelly, but it shouldn't make much difference.

A.

[Diabolical Artificer]

Off the subject of tfmr's I've been going through looking for obvious duff components. On the rectifier board I found a SC burnt out diode which I replaced with a 1N4007. Whilst I was at it, there are two PNP regulator trannys, one tests ok, tother tested ok BC at first, but tested SC CE negative probe on the emmiter, but tested ok apart from that.

I looked it up, it being a Motorola 1850-0098 - can't find a datasheet or JEDEC part #. So the Q next to it looks like a replacement, a MJ2267, I can't find much about this Q either. Went back and tested the 1850 etc, it now tests SC full stop, odd.

I should have some Mullard OC28's somewhere I could try, and maybe a few other PNP power trannies, but I'd be winging it. Any ideas?

Andy.

[orbanp1]

Hi Andy,

I can see you are making progress!

Looking still a bit more at the schematics, some more comments:

The +300V output is actually the sum of the +150V output, and another 150V regulator.
That means that the current on the "bottom" 150V regulator is the sum of the load on the +150V output and the +300V output!
So my previous statement that the RMS current is the same as the DC load current is only partially true!
Take this into account when you rewind the transformer!
But hopefully you can simply measure the original wires.

As for the PNP pass transistors they should be able to handle at least 150V Vce, or better have 200V Vce rating.
There are two transistor in cascade. The voltage is halved on them by R54 and R55 (at the bottom regulator), at the PS giving +200V. (The other two 150V regulators are similar.)
Q5 just "takes up" half the voltage, Q6 does the regulation.
When the PS is running there is not that much voltage on the transistors, when there is a short at the output the PS goes into current limit mode, that is when the whole raw voltage from the rectifying stage gets onto those two transistors, that is when the high voltage rating is important.
The raw voltage at the +200V stage is 265V, so a single transistor should be able to handle half of that safely, and so comes the 150V - 200V rating.
At the other two 150V regulators the max voltage to the pass transistors is a bit lower as the raw DC voltage is lower.

As for picking replacement transistors, Si transistors are perfectly all right, there is no need for Ge transistors. Actually, Si transistors are preferred, those are a lot more reliable than Ge transistors. The adjustment would take care the difference in the Vbe voltage from Ge to Si device.
That does not make life that much easier, TO3 transistors generally are no longer available. You can also use TO220 cased transistors, if you do not mind the not authentic look.
I looked at Digikey, the 2SA1668 would handle the load, it is a 200V 2A Si PNP transistor (do double check the current load), and those are not that expensive:
https://www.digikey.ca/products/en/d...=1&pageSize=25
I am sure similar transistors are available in the UK too.

An alternative is to scavenge TO3 Si transistors from older equipment.
Audio amps from the early '80s might have them, but might not have the high enough voltage rating, better check those.
Also, transistorized servo drives for machine tools (NC machines) have such transistors in abundance in them.
Though those usually have "house numbered" transistors in them, so you better know about their power supplies as well, but they were generally high voltage devices.
Techs usually saved such boards when the drives were scrapped. You could ask around.

Regards, Peter

[Radio Wrangler]

Originally Posted by Diabolical Artificer

I looked it up, it being a Motorola 1850-0098 - can't find a datasheet or JEDEC part #
Andy.

That's an HP stock number. It might be a plain JEDEC part bought in, or a Motorola proprietary type. The first four digita tell you what it is in general terms, the other 4 digits are just chronological.

1850 is germanium
1854 is silicon NPN
1853 is silicon PNP
1855 are all FETs
1901 are silicon diodes
1906 are multiple diodes

Look up HP part number crossreference on the web.

David
1902 are silicon zeners

[Humber888]

Andy,

Extracted from a June 2009 post:

Quote:

The 1850-0098 shows in Fedlog as:
PNP Junction
70V
7A
90W

Cross-Ref
B1493 - Solitron Devices
CQT794 - Silicon Transistor Corp
KT1117 - Silicon Transistor Corp
SP776 - Motorola
Also possibly a 2N301.

Fedlog is an acronym for the US Federal Logistics system.

[Diabolical Artificer]

Thanks Peter that helps a lot. The tfmr is wound oddly as far as I can make out. with the trouble I had extracting it, the wiring got a bit trashed. AFAICMO it's a split bobbin, two separate primary's, one on each side of the bobbin, with the secs split too.

I've made a replacement bobbin, got my coil winder out and re figured out how to use it, so am all set once I've done a few more calculation's.

From what I can make out, it will need two 150v @ 128mA + 50mA say to power the regulation circuit. So that would mean two 332v (2x 166) centre tapped at half 128 + 50, call it 200mA to be on the safe side, so one 332v @ 100mA. Working the 200v likewise gives me 510v (212+ 212 + 83) at 50mA The 6.3v has to supply two 8068's at 900mA each, call it 6.3v at 2A ish. I'll compare my calcs with what's on the old bobbin before I wind.

Talking of 8068's, both had loose top caps, one just fell off, so that's another problem that may have contributed to the demise of the tfmr on top of SC diodes and SC Q's. The big smothing can's check out ok with a quick ESR check, but I'll put each on a HV PSU and check for leakage.

I have a stash of old TO3 Q's, some off old TV's capable of up to 1500v, but also think I have some BUXXX, which AFAICR are HV types. So they should drop straight in? No changing the bias R's or anything?

Thanks for the info David and Humber, I think I saw a 2N301 yesterday whilst going through my TO3's.

Andy.

[orbanp1]

Hi Andy,

Did a quick DC simulation of the -200V circuit.

The current limit of the circuit is 58mA.

The current use of the regulator circuit is only about 18mA. (That would be the same for the other regulators as well.)
With a 50mA load (4k load resistor) the current draw from the 265V raw supply is about 68mA.

The current use from the -368V raw supply is about 16mA.
When in short circuit, from that same supply the current draw is 36mA!
While in short, the current draw from the 265V raw supply is about the same as without the short.

You do not want to overestimate the load by too much (safety factor), all the wire just might not fit onto the bobbin!

Transistors from the TV blocking oscillators are indeed a good source of high-voltage devices.
But be mindful of the transistor polarities though!
You need PNP transistors, and I would imagine that Si blocking oscillator transistors are all NPN!
I would also think that the BUxxx transistors are Si NPN devices!

Early TVs indeed used Ge PNP blocking oscillator transistors.
Some of those devices are:
AU106 Uce 320V Ic 10A
AU107 Uce 200V Ic 10A
AU108 Uce 100V Ic 10A
AU110 Uce 80V Ic 10A
AU113 Uce 250V Ic 10A

Regards, Peter

[Diabolical Artificer]

Hi Peter, thanks for the work you put in on the simulation, much appreciated.

Yep, Q's need to be PNP, wasn't thinking.

Have ordered some wire from Brocott. Measured the old tfmr and they used the same gauge wire for all the secs (apart from 6.3v that is) EG 30 guage.

Cheers, Andy.

[factory]

The 1850-0098 transistor is used in quite a lot of the HP oscilloscopes I have (e.g. 140A & 175A etc) from the 1960's, they did update these scopes during their production, the 140 series lasted the longest with the 140T & 141T spectrum analyser mainframes. Have a look at the circuit diagrams for these as some of the transistors did change to silicon types from the late 1960's onwards, probably still custom part numbers though and some circuits have been altered to used NPN types.

And if you get stuck I may be able to help with the original type if my spares are any good.

David

[factory]

The 140A used four 1850-0098 GE PNP transistors for the power supply series regulators in the earliest versions (1963), three of these circuits were changed to use SI NPN types by 1967, the remaining GE PNP changed to part no. 1850-0422, the 140T (1970) also uses the 1850-0422. These changes could definitely cause people trouble if they have the wrong year of manual and replace those transistors with the wrong type.

I've also had a look a later 1966 (paper) manual for the 175A and it still used seven of the GE PNP 1850-0098 transistors.

David

[factory]

Also had a look through the pictures on my PC and found my HP 140B (1973) uses a Motorola SJ 1798 in the place of the 1850-0422 GE PNP.
A web search for the SJ 1798 gives HP part number 1853-0252 and it's also a silicon PNP type.

David

[Diabolical Artificer]

Thanks David. The main bank of PNP Q's for the LV reg section appear to be well. The one I found dead was part of a pair mounted on the case side which is part of the LV rectifier board. I'll dig back into the paper work and found out what part of the circuit it's in.

Andy.

[Diabolical Artificer]

Quick update. I've wound the tfmr but it's been a hell of a job. the first bobbin I made fell to bits while winding due to my not paying attention and slacking off the wire tightness at the ends. So I made a bobbin out of copper clad board, soldered it up, great, nice and strong. Wound the two primarys and thought "you dope, the bobbin will be a shorted turn" Dremmel out, surgery done, shorted turn removed.

Wound on the secs, windings only just fitted. Checked voltages with part of the core on - the three secs were all shorted to one another....eh!!?? turns out the enamel had rubbed off where the wire goes from inside the bobbin, through a hole to the outside. Wire now shorted to the copper on the bobbin, hence 3 shorted secondary's.

I've now sorted that out and need to fit the core lams. Problem is they are a bit out of shape, this means they don't fit well and are tight. I've sat for hours straightening them out, all 300 odd and once fitted will check tfmr and test.

Really though I'll probably junk it and start from fresh - new bobbin, new lams, new wind.

Andy.