HP8568B calibrator noise / spurious signals

[Chris Wilson]

As a sanity check is it possible for you to look at TP17 on yours and check the ramp is different?

Q17 appears to be a 2N4117A. That's available from Little Diode.

Googling a X reference for HP part number 1854-0071 seems to suggest Q18 is a 2N3391, and some suggest a common of garden 2N2222 is also a suitable substitute. I have loads of those, should I remove Q18 and shove a 2N2222 in? There's also warnings of pin configuration differences for a 2N3391 / 2N2222, so I need to check that out. Would my semiconductor tester show issues with these 2, or are they probably "sort of all right" and only going faulty with fast signals? Cheers Jeremy!

[G0HZU_JMR]

I'm not sure if you have selected the correct device to change when you mention Q18.

I think the unity gain buffer is Q17 (N JFET) and Q19 (PNP). Sadly, the manual is a bit fuzzy so it isn't easy to tell which is which!

Q18 is used elsewhere in the constant current ranging circuit and is an NPN device.

I'll have a look at TP17 on mine and post up some plots. Note that you need to be careful when probing here. It may be possible to damage Q17 if you accidentally short TP17 to ground because it could damage the PN junction at the gate of Q17. But that's just a guess. The datasheet says this junction can be damaged at 50mA forward gate current and I'm not sure if the energy stored in C16 when charged to 5V could damage Q17 if you shorted TP17 to ground.

So best to be careful. I'm going to use a x10 probe at TP10. I expect it will be a very linear ramp.

[Chris Wilson]

OK, I see what's happened, David mentioned it as Q18, and he probably made a typo or his manual wasn't very clear, it is indeed Q19, and not Q18. I just looked up Q18 in the parts list first.

I can't see an equivalent, so have ordered a 2N3799 and a 2N4117A. If you know of a common equivalent for either let me know please! I can have a look in my box of bits.

Don't worry about probing if there's any risk at all, or it's a PITA, but if you feel happy it would be nice to compare the traces. Thanks Jeremy.

[Chris Wilson]

Just tested TP17 with the probe at X10 and see a smooth ramp...

[G0HZU_JMR]

It's a quick and easy job to tip it over on one side and remove the bottom cover so here's TP17 on mine.

The first plot is TP17 at the default 'bootup' 0-1500MHz sweep. The second is when the CF is 20MHz and the span is 1.005MHz.

I'm afraid I only have an old Tek TDS2012 scope so the display quality is pretty grim. But it shows a very linear ramp here. I also had a quick look at the -15V rail at the regulator board and it looks clean.

[G0HZU_JMR]

It looks like your x1 probe is loading the circuit I don't see any spikes in your latest plot of TP17. with the linear ramp. Have the spikes gone now you are using a hi Z probe? I'm guessing that they will still be there because they appear in so many places.

So we may still need to investigate Q17 and Q19.

[Radio Wrangler]

I misread the diagram (the one posted earlier in the thread) as it was rather blurry. Sorry for the confusion.

I really don't like X1 probes because of their very high unbuffered capacitance, but those ramps are 20ms which is slow enough that the charging current ought to be tolerable. Those transistors are a feedback pair and could exhibit stability problems with a capacitive load on the test pin.

Um, if those two pictures of the test pin waveform from Chris are both from the same point with X1 and X10 probes.... the flat pause at -5v becomes a non-linear ramp, then when the ramp proper begins the left trace ramp accelerates... um, this is a bit late to ask, Chris, was the left trace taken with the scope AC coupled or DC coupled?

David

[Chris Wilson]

All traces posted were taken AC coupled David.

Jeremy, many many thanks for investigating your own machine, that was very good of you. I can easily take DC coupled traces, the analyser is open for investigation. My probes are "supposedly" 200 MHz ones, but from China, via Ebay

[Chris Wilson]

Originally Posted by G0HZU_JMR

It looks like your x1 probe is loading the circuit I don't see any spikes in your latest plot of TP17. with the linear ramp. Have the spikes gone now you are using a hi Z probe? I'm guessing that they will still be there because they appear in so many places.

When i read your post re using a X10 probe I VERY quickly redid the traces as I had a party to go to with my wife. I just grabbed the first two traces that appeared. I will do some more tests tomorrow, the scope I have been using needs the "acquire" function used to see the spikes, and not every acquisition shows them, plus I need to be seeing spikes on the SA screen, usually, to see spikes on the test points.

[Radio Wrangler]

AH, that explains the bent waveforms. You were right, you did have two faults, but one of them could be fixed be setting your scope to "DC'

This should give you good linear ramps on the scope, and then we need to chase the spikes.

With DC coupling we'll be able to see what voltage the different spikes go to - this might be a clue.

I'm sorry it took me so long to twig that you were using AC coupling, it isn't often used, especially with slow or asymmetric waveforms. Treat DC coupling as the default for oscilloscopes and only use AC if you have an offset waveform you can't get onto the screen with the shift control when you are on the volts/div setting you want to use. AC is good for looking at noise on power supplies. It's useable for RF work as well, but sometimes throwing away the DC component leaves you late to notice that a bias point is badly wrong.

David

[Chris Wilson]

Hi David, no worries about the coupling, I was assuming the sikes were more noise than ripple and that AC was the better coupling.

I haven't had much time today, but I have noticed something odd, thanks in the main to playing with AC and DC coupling.

20 MHz centre frequency, with the calibrator signal fed in. 1 MHz span.

Scope on DC coupling, X10 probe setting


TP3 has a DC voltage offset. Over 1Mhz span (just over) the DC voltage offset is higher than in in an under 1 MHz span.

*Except occasionally* if you bring the span down on the rotary knob from just over 1MHz to just under 1 MHz, the DC offset will remain high....

Can anyone tell me if the DC offset SHOULD be changing? The waveform itself remains much the same.

Whether the DC offset tracks the change in span or not, the 2nd LO Hi LED works correctly and always comes on if the span is dropped under 1MHz.


I see the most noise on TP3.

TP2 (hidden underneath the multi pin connector on the top of A22), only gives much of an output on very wide spans, the schematic for it (YTO Main Coil Tune Driver) has "I.P." shown over the waveform drawing, and I am not sure what that means.

Other than the above I have not achieved anything worthwhile on it today.

I apologise for the terrible photos, I am not sure why some came out so badly!

The third photo is of TP17 with the scope set to DC and taken when the SA display was showing noise spikes.

[G0HZU_JMR]

Chris, I think you need to redo some of the plots above because I think you have some of the labels muddled up.

The plot with TP17 in the label can't be TP17 because the direction of the ramp is backwards.

The plot with TP15 in the label can't be TP15 because TP15 is meant to be a steady DC reference voltage for the ramp generator and not a square wave.

Can you do a plot of TP17 with DC coupling and a x10 probe with the scope set such that 0V DC is the centre line of the scope? Does it still show spikes here? This is an important test.

Also important... what is the TP number with the spiky square wave that you have labelled TP15? I can't see how it can be TP15 but I'm interested to find out which TP this is.

TP3 is the FM coil driver and it is driven by a summing amplifier U19 and Q24 and Q25. The summing amplifier is fed two inputs that it sums together. One input is the DAC attenuated sweep ramp (that is deferenced from the ramp at TP17)

The other input to the summing amplifier U19 is from the YTO PLL module (A6). This should be a DC voltage that corresponds to the tuning voltage obtained from the PLL. So TP3 should see the sum of a DC offset voltage and the attenuated ramp from the DAC.

[Radio Wrangler]

So now we're down to just the spikes looking wrond. The sweep generator now looks linear, as it should. You'v also got the lock and roll sequence where the sweep is taken to mid-screen and the PLLs are locked to correct the tuning to give the required centre frequency, then the sweep voltage jumps to the start of the ramp and does a linear ramp before jumping back to mid-screen.

I agree that TP17 (on the picture name atleast) looks inverted so is it TP17 with the scope invert activated? or was theprobe elsewhere?

David

[Chris Wilson]

I was under pressure today to do other things, I may well have muddled up the plot labelling, sorry, I took the shots then re-named them later. I can say for sure the square wave plot is taken at TP16.

I just grabbed some fresh shots before going to bed.

The machine hadn't been on long enough for the SA display to show any serious spikes, but nonetheless some spikes can be seen on the `scope display captures.

Hope these are more sensible! Thanks Jeremy.

[G0HZU_JMR]

Thanks Chris... I guessed that the square waveform could be at TP16 and this is the halt sweep section.

If so then it looks like it is getting spikes that correspond to false sweep stop/start signals. These signals ultimately come from the A12 module via the motherboard. A12 is next to the main controller board A15.

It's possible that Q26 is faulty (or Q2 is faulty on the A12 board) but I think there may be a more complex issue here to do with glitches in the system.
Eg the priority encoder may be dishing out illegal blips on the pin that controls Q2 on A12.


You could try monitoring activity on Q2 on the A12 module to look for spikes but I think we could be looking at some weirdness in the priority encoder chip U1 on Module A12. It could be caused by something as simple as a wayward pulse train from your front panel rotary encoder control. Or maybe it's the (global) phase lock input flag to the priority encoder on Module A12 chip U1 pin 7 that is intermittently reporting a phase lock fault somewhere in one of the system PLLs ?

I'm afraid it's nearly bedtime and my head can't think too clearly so maybe revisit all this tomorrow

[G0HZU_JMR]

It could be that the weird glitch activity at the priority encoder A12 U1 is caused by a very brief PLL unlock flag at A12 U1 pin 7 that is initially caused by a fault on the A22 board. eg something to do with the ramp originally generated at TP17. So there could be a chain reaction. Otherwise I can't reliably explain why there are big spikes at both TP16 and at TP17.

So we could be back to looking around Q17 and Q19 on A22 for the initial cause of all this spiky mayhem ocurring at TP17, TP10, TP3, TP16, the -15V rail and even A6 TP4...

[Chris Wilson]

OK, thanks Jeremy. I took a week away from it, work to do to keep food in the fridge and HMRC happy Also a new puppy to remind me of how time consuming and demanding they are at first.

So I will return to it. The two transistors for Q17 and Q19 arrived, I'll just fit them, but also test the removed devices.

On A12 spraying freeze spray on U1 and Q2 had no effect. A brief squirt around the optical encoder area behind the front panel may have had some effect. The noise comes and goes so seemingly randomly it's easy to fool yourself that something you just did had an effect, when the change may have happened anyway. Are there any useful test points on A12 to look at? Without an extender board I would need to solder short wires on as they are deep down on the board.

Is there a way to disconnect the optical encoder and still have the SA run, in order to see if it;'s giving random signals? It operates smoothly and I see no obvious jumps when turned really slowly.

[Radio Wrangler]

The rotary pulse generator, 'RPG', ought to be OK. You can unplug it and just type frequencies in. Anyway, I think I have a few spare RPGs

David

[Chris Wilson]

Thanks David, I see no obvious way to disconnect it save un-soldering the pin connectors on the front PCB.

I have changed Q17 and Q19 for new devices, but the noise is the same.

Please humour me, I have a gut feeling that the issue is around R39. If I solder a wire on to the hard to get at TP2 (YTO Main Coil Tune Driver) on A22, and touch the end of the wire, the calibrator signal jumps side to side all the while I am touching it. If I touch and let go a few times the noise spikes start in a dramatic fashion very quickly. I have posted before that at times just brushing the adjuster screw of R39 when the jitter (not noise spikes necessarily) is bad, makes it far worse. I have renewed R39 to no avail. It's as if this part of the circuit goes wild, a bit like when something with an adjuster pot is not biased right and the pot is far far too "touchy". is there anything about this circuit that could be changing and causing instability? Anything I can add or remove signal wise, voltage wise or connection wise around there to isolate the coil driver somehow?

In my post #131 (blinkin' `eck, this is a big thread...), I mentioned the below, is it significant, because it is repeatable:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~~~~

"I have noticed something odd, thanks in the main to playing with AC and DC coupling.

20 MHz centre frequency, with the calibrator signal fed in. 1 MHz span.

Scope on DC coupling, X10 probe setting

TP3 has a DC voltage offset. Over 1Mhz span (just over) the DC voltage offset is higher than in in an under 1 MHz span.

*Except occasionally* if you bring the span down on the rotary knob from just over 1MHz to just under 1 MHz, the DC offset will remain high....

Can anyone tell me if the DC offset SHOULD be changing? The waveform itself remains much the same.

Whether the DC offset tracks the change in span or not, the 2nd LO Hi LED works correctly and always comes on if the span is dropped under 1MHz.

I see the most noise on TP3.

TP2 (hidden underneath the multi pin connector on the top of A22), only gives much of an output on very wide spans, the schematic for it (YTO Main Coil Tune Driver) has "I.P." shown over the waveform drawing, and I am not sure what that means."
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ~

In particular, is this occasional variance from a DC offset change to no change normal and expected, as it seems random? I can't work out why it should occasionally do something different in response to a span change, I would have expected it to either have a DC offset change or not, each time at a span change, every time.

Thanks.

[G0HZU_JMR]

(YTO Main Coil Tune Driver) has "I.P." shown over the waveform drawing, and I am not sure what that means."

IP means Instrument Preset so that waveform refers to the default 1500MHz span at powerup or if you press the green PRESET button.

Can you check pin 7 on the U1 chip on A12 to see if there is activity here from the global PLL unlock flag when there are lots of noise spikes?

This is much more likely that the RPG. I only gave the RPG as a 'possible' cause of spurious activity here because it also feeds to the priority encoder U1 on A12.

The global PLL flag at A12 U1 pin 7 has various status 'triggers' that cause it to show an unlock flag.

These include:

Main YTO PLL unlocked
249MHz loop unlocked
YTO tune voltage LOW
YTO Tune voltage HIGH
275MHz Loop unlocked
Main OCXO Oven Cold

There could be a chain of events that happen that cause spikes to appear in lots of places. I think the spikes at TP16 on A22 may be caused by brief pulses on the the PLL unlock flag.

So this would just be fallout from the 'real' fault that could be elsewhere on A22.

Do you see spikes on the global lock flag on U1 pin 7 on the A12 module? I think it is worth ruling this in or out as part of the fault condition.