EEVblog Electronics Community Forum

Electronics => Repair => Topic started by: grumpydoc on July 16, 2017, 07:35:18 pm

Title: Fluke/Phillips PM3394 "Delta-T" repair (Help with circuit interpretation)
Post by: grumpydoc on July 16, 2017, 07:35:18 pm
What seems like ages ago (sadly I keep getting interrupted by other stuff) I agreed to have a look at the CPU and digitiser board from a Philips/Fluke PM3394 "combiscope" for another forum member (EHT).

I have the CPU card running - the problem having been corrupt firmware, however before I sent the cards back I thought that it would be a good idea to check the A8 board - here there was more trouble.

The digitiser needs to be able to measure 10-20ns pulse lengths down to a few 10's of ps precision but only has a 200MHz clock so uses an analogue circuit which Philips named the "Delta-T" circuit in the manuals. Sadly this is not working in EHT's board.

Basically it works by charging a capacitor via a constant current source during the pulse to be measured, then discharges it via another constant current source but at approx 1/500th the rate - by timing the discharge against the 200MHz clock we get resolution down to 10ps or so.

In EHT's board the cap does not appear to be discharging.

Now in later versions of the board (after about the end of '92) the circuit looks like this and I think I can follow it well enough.

(http://www.wild-pc.co.uk/images/PM3394-9207-p440.jpg) (http://www.wild-pc.co.uk/images/PM3394-9303-p440.pdf)

Click on the image for the pdf of the full page in the service manual for extra context.

So, R8606 & 860& form a divider from the 10V rail to give a 5V reference. Op amp N8074 will keep the junction of R8489 and the emitter of V8127 at the reference voltage - giving 3.086mA through R8489 - consequently the voltage across R8493 will be 3.086V, the voltage across R8496 will also be 3.086V so 60uA will flow through R8453/V8136/V8134/R8496.

this 60uA is the discharge current, the charge current is generated in an (essentially) identically configured pair of op amps (see PDF) but with resistor values chosen to give a charge current of approx 30ma. The two BFT93 transistors are used to steer the charge current either to the integrating capacitor or down to ground via R8449.

The only thing I am a bit puzzled about is the role of V8136.

However in the version of the A8 board that EHT sent me the circuit is a bit different.

(http://www.wild-pc.co.uk/images/PM3394-9207-p311.jpg) (http://www.wild-pc.co.uk/images/PM3394-9207-p311.pdf)

I presume that the current through R8489 and R8493 are supposed to be the same (as the later circuit) but I don't fully understand how it works - in particular the role of V8207 in the feedback loop.

Can anyone help out in explaining the operation of the circuit?
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 16, 2017, 07:41:00 pm
Is that the miller effect compensator.?

https://www.youtube.com/watch?v=Op_I3Ke7px0 (https://www.youtube.com/watch?v=Op_I3Ke7px0)

was [26027 views]
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 16, 2017, 07:57:15 pm
Is that the miller effect compensator.?
Seems unlikely as this circuit is basically operating at DC.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 16, 2017, 08:01:45 pm
Is that the miller effect compensator.?
Seems unlikely as this circuit is basically operating at DC.

You don't need a choke [inductor] if it's purely DC!
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 16, 2017, 08:11:22 pm
Is that the miller effect compensator.?
Seems unlikely as this circuit is basically operating at DC.

You don't need a choke [inductor] if it's purely DC!
That will be for additional filtering of any noise on the 18V rail - that inductor passes the 30mA charge current but no AC signal at all. The circuit is arranged so that the 30mA flows continuously - you could not turn the charge current on and off with the pulse to be measured as it is way too short (as I said only 10-20ns) - the control loop would never settle in time.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 16, 2017, 08:24:00 pm
Yeah! maybe its an extra component, you can try take it out to see the effect.  :-//
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 16, 2017, 08:32:44 pm
Yeah! maybe its an extra component, you can try take it out to see the effect.  :-//
No, the plan is to measure the DC voltages around the current sources and see if there is an obvious fault. I tried measuring the current steering pulses but it is impossible to get short enough probe leads onto the PCB with the 'scope running - however, except for a short period where the cap is supposed to be charging it's largely DC conditions, and - as I explained - the current sources are intended to operate continuously so all the voltages around them are DC anyway. It's just in a very awkward location on the PCB to probe.

However I would like to properly understand how the current source operates.

Edit: Your comment is rather flippant and implies you know but consider it beneath your elevated expertise  to explain. Which is hardly fair as I had (I thought) politely asked for help.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: The Soulman on July 16, 2017, 08:35:12 pm
Don't worry about the choke or the additional diode to much.
Have you looked at V8126 and the control signals RUNDT and RUNDTN?

Edit:


No, the plan is to measure the DC voltages around the current sources and see if there is an obvious fault. I tried measuring the current steering pulses but it is impossible to get short enough probe leads onto the PCB with the 'scope running -

Ok , can you probe across R8449?
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 16, 2017, 08:49:06 pm

Edit: Your comment is rather flippant and implies you know but consider it beneath your elevated expertise  to explain. Which is hardly fair as I had (I thought) politely asked for help.

Hi, cheers!, I really don't know because I have not spend a great among of time reading the circuit. Relax, someone might know and offer you a solution, OK?.  :-+
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 16, 2017, 08:50:39 pm
Don't worry about the choke or the additional diode to much.
Have you looked at V8126 and the control signals RUNDT and RUNDTN?
I'm certainly not worried about the choke :)

As I said my first thought was to look at RUNDT and RUNDTN and make sure that they were present.

There are a couple of problems however - all this stuff is on the bottom edge of the digitiser card. With the card in the 'scope you have about 3cm one side and 4cm the other and about 10cm down so it is impossible to get a 'scope probe onto the board. I don't have an extender card which would be very useful.

I managed to get a logic probe clip onto RUNDTN but could not see a pulse - however that is probably due to the length of the ground loop as I could not get the ground lead anywhere near the point of measurement.

The 2nd issue is that I'm not even sure what levels the signals should be - I don't think that I trust the 8/8.5V labels (there are definite mistakes elsewhere - there is not, for instance, 3.5mV across R8491, it can easily be seen that it should be about 3V). The signals originate in a custom chip which is combined ECL and TTL logic and fed with -4.5V and 5V supplies so I can't see that it can generate 8.5V on its outputs. They could be open collector/drain I suppose but even then the limit would be the 8.3V generated by V8114.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 16, 2017, 08:54:57 pm

Edit: Your comment is rather flippant and implies you know but consider it beneath your elevated expertise  to explain. Which is hardly fair as I had (I thought) politely asked for help.

Hi, cheers!, I really don't know because I have not spend a great among of time reading the circuit. Relax, someone might know and offer you a solution, OK?.  :-+
Sorry, too many conversations recently where people have been critical but offered no useful insight (not on the forum BTW). Guess I got a bit more sensitive to that than I thought.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Andy Watson on July 16, 2017, 09:11:25 pm
I think there may be a bit of misunderstanding about the "choke" - I think what Armadillo is saying, and I agree with it, is that V8136 is indeed operating in cascode mode and is intended to isolate the the high-frequency pulse on its drain from the rest of the discharge circuit. As you say, the circuit that defines the 60uA is operating at DC, but if you were to disturb it, it will only settle as fast as the bandwidth of the op-amps will allow - which is presumably much less than the bandwidth required to react to the pulse on the drain of V8136.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 16, 2017, 09:32:22 pm
I think there may be a bit of misunderstanding about the "choke" - I think what Armadillo is saying, and I agree with it, is that V8136 is indeed operating in cascode mode and is intended to isolate the the high-frequency pulse on its drain from the rest of the discharge circuit. As you say, the circuit that defines the 60uA is operating at DC, but if you were to disturb it, it will only settle as fast as the bandwidth of the op-amps will allow - which is presumably much less than the bandwidth required to react to the pulse on the drain of V8136.
Ah, thanks, that makes some sense. Totally missed the fact that Armadillo was talking about V8136 and perhaps a little too focussed on the DC conditions.

The business end of the discharge source doesn't really operate at DC, thinking about it - the duty cycle is pretty low so the cap is mostly discharged (or would be if things were working) so it will spend most of the time not passing any current and have to settle pretty quickly once the cap starts to charge..

I remain puzzled as to exactly how the 2nd (earlier) version of the circuit regulates the current through R8489/R8493.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Andy Watson on July 16, 2017, 10:02:33 pm
I remain puzzled as to exactly how the 2nd (earlier) version of the circuit regulates the current through R8489/R8493.
Both versions of the circuit are enforcing a constant voltage across R8489 (note that the inputs to N8074 have been reversed) - so the output is a constant current which drives R8493. It's a bit odd that they have used this configuration of one current source providing the reference for the other current source/mirror  - so I'm guessing that they consider the 10VDT power supply to be the most stable voltage reference.

Edit: Also, they are using the 10VDT supply as the reference for the charging current source - so even if the 10V is not exact, the charge and discharge currents will track each other. The odd configuration is simply a method of transferring the positive reference to the negative supply - as required by the discharge source.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 16, 2017, 10:24:50 pm
I remain puzzled as to exactly how the 2nd (earlier) version of the circuit regulates the current through R8489/R8493.
Both versions of the circuit are enforcing a constant voltage across R8489 (note that the inputs to N8074 have been reversed) - so the output is a constant current which drives R8493.
Any chance of expanding how the control loop works - I know I'm being dense but I can't quite figure it out.

Disregarding the diode for a sec you have the 5V reference on the inverting input and the sense on the end of R8489 - if the current drops then the voltage on the non-inverting input will rise and the op amp output will start to swing up to 18V, that will allow more current to flow through the FET, restoring the drop across the resistor. Conversely if the curren increases, the voltage at the non-inverting input will drop below the inverting input and the output will head negative cutting the current through the FET.  So I think I get the basic loop. But why the diode feedback. Assuming that the output of the 358 is less than 0.6V above the reference the diode won't be conducting, but if it is then it will, presumably, conduct and shift the reference voltage so I don't get the detail of what that's suppose to achieve.

Quote
It's a bit odd that they have used this configuration of one current source providing the reference for the other current source/mirror  - so I'm guessing that they consider the 10VDT power supply to be the most stable voltage reference.
It probably is - the 10V line is regulated from the 12V line using the 'scope's primary 10V reference.

Quote
Edit: Also, they are using the 10VDT supply as the reference for the charging current source - so even if the 10V is not exact, the charge and discharge currents will track each other. The odd configuration is simply a method of transferring the positive reference to the negative supply - as required by the discharge source.
OK

Actually nothing in this circuit needs to be exact, just stable. Part of the autocal procedure is to time a 10ns pulse, then a 20ns pulse. That will give it two points to work out a scale factor & offset when doing the measurements.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Andy Watson on July 16, 2017, 10:55:57 pm
I think your explanation for the FET version is correct. The bipolar version is probably best viewed as a voltage follower aimed at fixing 5V across R8489.

I was going to suggest that the diode feedback was there to prevent some kind of latch-up situation during power-up. Remember that these are jfets - if you forward bias the gate-source it is a PN junction which will conduct. Also note that the op-amp has positive feed back from R8489. If the output of the op-amp were sufficiently high, it is possible that it could supply all the current to drive the R8493 via the PN junction of the jfet - so R8489 is robbed of current and the positive feedback maintains this situation. The feedback diode checks the voltage on the output of the op-amp.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 17, 2017, 05:20:53 am

In EHT's board the cap does not appear to be discharging.


How did you test that the capacitor C8271 is not charging/discharging? Please explain.

I don't suppose you actually measured the voltage from the poor 150pf capacitor?

Can you confirm that pulse train is coming into base of V8126 and then if N8062 Pin 4 [DTRDYN] has a pulse train out, C8271 is charging and discharging has been confirmed. Can you confirm this?

Don't fix your thinking into the DC opamp theory too much, because this circuit is pulsed On/Off very rapidly, hence the dynamic transient response, capacitive effect on FET gate comes into play, e.g. to isolate the important key capacitor C8271 from drain-gate capacitance of V8134 [60uA source] so as not to affect the precision current source and glitch free.

Let us know your findings.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 17, 2017, 02:15:35 pm

In EHT's board the cap does not appear to be discharging.


How did you test that the capacitor C8271 is not charging/discharging? Please explain.

I don't suppose you actually measured the voltage from the poor 150pf capacitor?
Well, I stuck a 'scope probe on it looking for the sawtooth. It was just showing about 8VDC (need to repeat that to re-check the exact voltage but definitely no sawtooth).

I have confirmed that, on a working board, the sawtooth is visible without too much difficulty (on a Lecroy 9354).

Quote
Can you confirm that pulse train is coming into base of V8126 and then if N8062 Pin 4 [DTRDYN] has a pulse train out, C8271 is charging and discharging has been confirmed. Can you confirm this?
Well, I think I can confirm that with a trace displayed (so the 'scope is triggering) I can't see a sawtooth on C8271.

My next thought was to look at the bases of V8126 and V8124, I managed to get a probe hook on the base of V8124 but could not see a pulse there. However because of access difficulties I had long leads, especialy the ground clip which was attached "miles" away. So a 20ns pulse might be difficult to see. Also it could be quite a small swing that I am looking for - the 8/8.5V label suggests that it might only swing through half a volt.

Quote
Don't fix your thinking into the DC opamp theory too much, because this circuit is pulsed On/Off very rapidly, hence the dynamic transient response, capacitive effect on FET gate comes into play, e.g. to isolate the important key capacitor C8271 from drain-gate capacitance of V8134 [60uA source] so as not to affect the precision current source and glitch free.

Let us know your findings.
My current thinking is this:

I can't demonstrate that the RUNDT/RUNDTN pulses are good. If they are not then the board is scrap as they originate in one of the large custom chips and they connect almost directly to the bases of V8124/6. But let's assume for a moment that they are OK.

At some point I will swap back in a known good card and see if I can see those pulses on a 'scope - I'll know what I am looking for at that point and know that I can probe it so I can see if it is present on the faulty card. In the mean time I was trying to find a 20ns pulse of unknown amplitude with a 15cm unscreened lead and about 20cm of ground loop. I'm not terribly surprised that it elluded me!

I want to limit board swaps though - EHT's card is pretty much "1st generation" and has a lot of (factory) re-work on the card so it is a bit mechanically fragile, some of the RF connectors on the board need quite a bit of pulling to get off and I don't want to keep doing that repeatedly.

So, given that the charge pulse is short, the charge source runs all the time and is steered to and from the integrating capacitor and the discharge source is connected all the time then there are a few things that I can look at with DC levels.

I can look at VBE on V8124/6 and make sure that V8124 looks as though it is not conducting and V8126 looks as though it is.

I can look at the voltage on R8496 and see if the discharge current is flowing - if not (which I suspect/hope) I can check the voltages around V8074 (which is definitely just operating at DC) and see if they look OK (but for that I need to know what they should be as it isn't given in the manual - hence this thread).

If the voltages around N8074 and the reference input to N8066 look OK but there is no current flowing (which there should be if the cap is charged to 8V) then I'll probably replace N8066/V8134/V8136 and check their associated resistors and see where that gets me

Sound reasonable?

Not sure when I'll get any time on the workbench though. "Real" work is a bit busy at the moment.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 17, 2017, 02:48:25 pm
I am sure you very well know what you are doing.

For me, however,

If I am not able to ;

1) Get any long pulses from DTRDYN, then I would confirm
2) Any pulses on the RUNDT;
3) If I can't get the confirmation in item [2] then

I would be looking else where example the clock oscillator circuit - MCAA clock, WHISTLER circuit etc.  I would not be spending time on the DC voltages knowing that without the pulses the charge / discharge would not be initiated at all.

And finally accept the demise of the jungle IC chip and move on forward to next venture.

Sound reasonable? No.... check elsewhere until finally..... you decide.

 
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 17, 2017, 03:24:34 pm
All very reasonable points. Note that the 'scope will show a mostly stable waveform in digital mode up to the 250us/div timebase (faster than that and it switches to equivalent sampling and lack of a working delta-t screws it at that point, prior to that you just get a little jitter and a delta-t timeout message).

So clock, ADC, DSP, RAM and triggering etc are working. As above the 'scope is confirming it is not seeing the end of slope signal with a "Delta-T timeout" error message.

I chose to start with the integration waveform as it is the mid point of the Delta-T circuit - as there is no sawtooth there there's no point looking at the buffer or comparator. I agree 100% that checking that the RUNDT/RUNDTN pulses is the logical next thing to do but have hit some practical difficulties with that.

I'll get back to that fairly shortly but in the interim want to check the analogue bits on the charge/discharge sources to see if I get lucky or can infer something about the RUNDT/N signals with an easier measurement.

At the moment I reckon that the drop across R8496 is the key, easy measurement - if the top of C8127 is at 8V there should be 60uA flowing through R8496 and about 3V across it. If so the problem is RUNDT/N or the steering transistors. If not the problem is possibly with the discharge current source (which is actually what I'm hoping).

tl;dr - yes agree 100% in theory regarding the next logical step in testing. In practice i'm going to take a short detour first.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 17, 2017, 03:29:41 pm

How did you test that the capacitor C8271 is not charging/discharging? Please explain.

I don't suppose you actually measured the voltage from the poor 150pf capacitor?

Well, I stuck a 'scope probe on it looking for the sawtooth. It was just showing about 8VDC (need to repeat that to re-check the exact voltage but definitely no sawtooth).

I have confirmed that, on a working board, the sawtooth is visible without too much difficulty (on a Lecroy 9354).

This one really puzzled me.

You know that the charge time is max 20ns @ 30mA, so the charge is very very small. How would you rationally put a probe [full of capacitance] to rob the charge away from the poor 150pf capacitor C8271. You at least should be measuring it after the buffer V8137 if you really wanted.

Furthermore, assuming maximum charge time of 20ns, the max voltage charged on the capacitor would be only 3.29682635266 V, how did you managed to confirm 8v and is able to measure a saw tooth waveform there from a known good board you said??, really puzzled me. Furthermore, it is clearly written on the schematic 0....3v SAWTOOTH.

Do a exercise;
Capacitor 150pf,
Voltage 10V
Series Resistor 333.33 ohm [so the charge current is 30mA]
Time of charge 20nsec
What is the charge voltage? 3.29682635266 V which somewhat tally with the schematic.

Now, conversely speaking, if you really measured 8V, then something is wrong with those transistors and it is worth to investigate the transistors there.




Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 17, 2017, 03:37:35 pm
The sawtooth on the good board was not 8V, it was about 3V, as expected - sorry if that was not clear.

If the faulty board was charging the cap would the voltage not rise higher over several measurement attempts, given that it probably isn't being discharged correctly?

At least that is my working assumption.

PS: Yes, puttng a probe at the top of the cap is certainly going to add capacitive and resistive load and alter the behaviour of the ciruit - but not enough to stop seeing the waveform at all. That was one of the reasons I checked on a known working board.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 17, 2017, 03:53:37 pm
The sawtooth on the good board was not 8V, it was about 3V, as expected - sorry if that was not clear.

If the faulty board was charging the cap would the voltage not rise higher over several measurement attempts, given that it probably isn't being discharged correctly?

At least that is my working assumption.

PS: Yes, puttng a probe at the top of the cap is certainly going to add capacitive and resistive load and alter the behaviour of the ciruit - but not enough to stop seeing the waveform at all. That was one of the reasons I checked on a known working board.

You have a lot of reasons........ LOL...  |O
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 17, 2017, 09:07:39 pm
I can now confirm that RUNDT/RUNDTN are present and the voltages correspond to those in the schematic "1" to "0" swing is just under 0.5V and the actual voltages seem to be about 7.9 and 8.3V which, I think, confirms my suspicion that the 8.5V in the later manual is not quite right as the highest local rail is 8.3V, nor is the 7.5V in the earlier manual which is possibly why they changed it.

It was a slightly difficult signal to capture - I managed to raise the digital board out of the 'scope on some stacked DIN 41612 connectors so I could get a probe on the two 31R6 resistors on the bases of V8124/6. I was still stuck with a 10cm ground clip but could now get it onto the central "ground" tab that normally anchors the board into the 'scope, which is right by where the two transistors are located. Then the trick then was to use a single shot capture on the LeCroy. One of the other annoying features of the signal is that the duty cycle is low because the waveform update/trigger rate on the PM3394 is very slow. I'd have never seen it on an analogue 'scope (or my 3394B for that matter).

I need to look at the voltage on the cap again - I don't seem to be able to reproduce the earlier 8V measurement.

I get 3.08V across R8496 (on a multimeter) which is puzzling me, mainly because I get 3.07V on the "good" board and AFAICS the cap voltage should tend to 0V for most of the time, so how can current flow in the discharge source.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 17, 2017, 10:04:13 pm
I get 3.08V across R8496 (on a multimeter) which is puzzling me, mainly because I get 3.07V on the "good" board and AFAICS the cap voltage should tend to 0V for most of the time, so how can current flow in the discharge source.

What cap are you talking about - ?

Exercise; [Proposal]

Separate circuit and measure. Please confirm findings; See diagram;

Lift up one leg of the inductor and the resistor or whichever, just apply concept.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: EHT on July 17, 2017, 11:15:13 pm
Firstly, thank you again GrumpyDoc and everyone else who has contributed on this thread, as it's my board!

I can now confirm that RUNDT/RUNDTN are present and the voltages correspond to those in the schematic "1" to "0" swing

 :clap: Excellent news! It lives!

This is all excellent work but I'm having trouble keeping track of the measurements. GrumpyDoc - have you tried checking for short circuits on the opamps & FETs & diode test on V8207 & V8124? This might also reveal if there is a mechanically caused short/open on the PCB. I don't know if this is one of the areas with rework but it all looked very fragile.

I wonder why they changed V8127 from the FET & diode to bipolar? Must be some reason, which perhaps could have been poor reliability somehow? The diode looks like a protection which limits V8127 G to approx 5.5v.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 23, 2017, 08:17:29 pm
Sorry for the lack of updates but I got diverted by the presence of new toys :)

Armadillo - sorry didn't quite do the measurements you suggested. I did measure the voltage across R8491 and got the expected 3.08V so the charge current source was running. I then had another look at the voltages around V8124 and V8126 - this time with a DMM rather than the 'scope to get accurate numbers rather than a general impression and they looked a bit odd - in particular the BE voltage on V8124 was too low.

Removed the two transistors and sure enough V8124 is shorted. Replaced them with ones taken from a dead digitiser board and bingo! no more "Delta-T" timeout messages and the trace actually looks OK in equivalent time mode. A bit noisy but I'm hoping that will go away with an autocal so the 'scope has the correct constants to use for time measurements.

Looks like it might be fixed!!  :-+

I'll run it through an autocal tomorrow and see if it passes.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: Armadillo on July 23, 2017, 08:43:31 pm
Great you done it. Congrates, good to know that you nailed it.

 :-+

All the best tomorrow.
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: EHT on July 24, 2017, 05:14:27 pm
 :clap:

Thanks GrumpyDoc and those who contributed to the thread!

I'm so glad it wasn't one of those ASICs!
Title: Re: Help with circuit interpretation (Fluke/Phillips PM3394)
Post by: grumpydoc on July 24, 2017, 08:49:24 pm
:clap:

Thanks GrumpyDoc and those who contributed to the thread!
Glad to have helped - I learned some bits, which is always worthwhile.

Quote
I'm so glad it wasn't one of those ASICs!
Indeed.

I am pleased to say that the 'scope passes its autocal procedure with the repaired digitiser card in situ so there shouldn't be any other faults on the board. Oh, I got the two transistors mixed up - it was V8126 which was shorted. Not sure how I got the 8V measurement though across the timing capacitor, I think I mentally applied the 10x for the probe when I had already set the 'scope for a 10x probe attenuation (thus multiplying by 100, not by 10). It lead me astray a bit in looking at the discharge current source first but I got there in the end :)

To add a few photos to the thread this is what the trace looked like originally when the 'scope was trying to use the equivalent time sampling mode to display a 2MHz sine wave.  :-BROKE

(http://www.wild-pc.co.uk/images/IMG_8538.jpg) (http://www.wild-pc.co.uk/images/IMG_8538.jpg)

The same input now looks like this  :)  :-+  :clap:

(http://www.wild-pc.co.uk/images/IMG_8600.jpg) (http://www.wild-pc.co.uk/images/IMG_8600.jpg)

Here is the sawtooth waveform on the timing capacitor (C8271) measured on another digitiser board. The duration seems to be about 7us so the original pulse must have been about 14ns. Sorry about the wobbly camera work!

(http://www.wild-pc.co.uk/images/IMG_8541.jpg) (http://www.wild-pc.co.uk/images/IMG_8541.jpg)

Finally I couldn't let the board go without some photos of the re-work that has been done. This was actually done by Phillips and is documented in the service manual but I had not seen it on a board before. In fact I have not seen this degree of rework on any commercial board ever and there is much more of it than shown in just these two shots. There are whole rows of transistors tagged on top of other transistors (a few can be seen above and to the right of the green wire in the first shot.

(http://www.wild-pc.co.uk/images/IMG_8603.jpg) (http://www.wild-pc.co.uk/images/IMG_8603.jpg)

(http://www.wild-pc.co.uk/images/IMG_8604.jpg) (http://www.wild-pc.co.uk/images/IMG_8604.jpg)
Title: Re: Fluke/Phillips PM3394 "Delta-T" repair (Help with circuit interpretation)
Post by: tautech on July 24, 2017, 08:59:17 pm
Good work grumpydoc, how long has this scope tormented you.....a couple of years at least IIRC.

Will it go down the road now you have that 1 GHz LeCroy ?
Title: Re: Fluke/Phillips PM3394 "Delta-T" repair (Help with circuit interpretation)
Post by: The Soulman on July 24, 2017, 09:01:03 pm
Good news and thanks for reporting back!  :-+

It look likes you use a camera with a 1.8 (ish) lens?  :-/O
Title: Re: Fluke/Phillips PM3394 "Delta-T" repair (Help with circuit interpretation)
Post by: grumpydoc on July 24, 2017, 09:28:52 pm
Good work grumpydoc, how long has this scope tormented you.....a couple of years at least IIRC.
Thankfully it hasn't all been the same 'scope!

I started with a PM3382, quickly decided that I like them as oscilloscopes but wanted a 3394B - which I sort of gradually acquired in bits of mostly non-working ones.

I have 5 working out of 7 purchases which isn't that bad :)

I really need to sell the spare ones on eBay though.

In fact this repair isn't one of mine - I offered to help another forum member out after he bought one on eBay which  the vendor was convinced had a "small PSU fault". With these things it is *much* easier if you have some working exemplars from which to take notes and measurements.

Quote
Will it go down the road now you have that 1 GHz LeCroy ?

Probably not - I usually turn the Phillips on first and most of the time it is all I need. The fans on the LeCroy are a bit loud for my taste but it will be useful when I need something with deep sample memories, high bandwidth and/or the more versatile triggering options. It should be easier to do screen shots on the new 'scope as well so I think the WavePro will see more use than the 9354 did.

What I will definitely now sell is my "spare" 9354 (I bought two for £125 a few years ago) and I really ought to get at least some of the spare PM33xx's calibrated and on eBay.

Good news and thanks for reporting back!  :-+

It look likes you use a camera with a 1.8 (ish) lens?  :-/O
I have a couple of lenses - A Tamron 10-24mm and a Canon 18-200mm, I'd quite like something as fast as a f1.8 as neither of the two are better than f3.5 and the workshop1 does not have much natural daylight so I usually tend to be using ISO1600-3200 and trying to hand hold 1/20th of a second. Image stabilisation helps a lot of course but I get quite a few shots which look like the trace on the LeCroy above.

The shots of the PCB were done with the 18-200 through a x3 macro lens, hence the shallow depth of field.

EDIT:
[1] Well, I say "workshop". In reality it is a bench along one wall of the garage amidst ever higher piles of stuff which I need to either fix or admit I can't and get rid of. As someone said in another thread "hoarding is part of the hobby" :)