EIP 545A 18GHz counter questions. Signature analysers?

[G0HZU_JMR]

Oh I see... sorry I didn't realise you measured it OK before you did the changes.

It is definitely messing with us

[Chris Wilson]

For completeness, (and to learn a bit more about signal tracing with my scope) I have also now probed the anode of CR3 and as you said, there's a much bigger signal level after the amplification.

http://www.gatesgarth.com/scope/CR3anode.jpg

and with 500MHz in at -10dBm:

http://www.gatesgarth.com/scope/500MHz.jpg


and with 1GHz in at -10dBm:

http://www.gatesgarth.com/scope/1GHz.jpg

[Chris Wilson]

I did a further test using the EIP testing trees, by looking for the TTL signal on TP12. Now, no having looked at TTL signals before, I can only set my scope up to see the voltage change, I know I could possibly see the waveform itself if I knew what I was doing. But Here is the test EIP suggest:

http://www.gatesgarth.com/scope/band-2-test.jpg


With no 50 MHz -20dBm input:

http://www.gatesgarth.com/scope/no-input.jpg

With 50 MHz -20dBm into Band 2 input socket:

http://www.gatesgarth.com/scope/50MHz-in.jpg

[G0HZU_JMR]

That's a test of the comparator action on Range 2A (10-185MHz)

It's a top level test of the signal path from RF to IF on range 2A with a -20dBm 50MHz signal.

The comparator U1 compares the signal strength at both the RF input and the IF output.

You can see it slightly reverse biases the IF diode CR3 from the -5.2V rail via a 160k resistor. This makes this diode slightly deafer than CR1 on weak signals or noise.

This is presumably to make sure the circuit behaves if there is no signal fed at the input and to prevent CR3 detecting low level noise which would falsely trigger a fake result at TP12.



So on a healthy counter the -20dBm 50MHz signal will be enough to be detected by CR1 and this sets a DC voltage at U1 pin 2. The bigger the signal, the bigger the DC voltage.

So this DC voltage sets the 'target' at the comparator for the IF detector CR3 to beat.

Basically in range 2A a healthy counter has to bias on the mixer MX1 using Q12 so it becomes a low loss bypass. The signal has to get through Q1, MX1, the 200MHz LPF and the amp Q8, Q9 and arrive at CR3 'big' enough to be detected to a DC voltage at comparator U1 pin 3 that is bigger than the DC voltage at U2 pin 2. If this happens the comparator will flip state to a logic 1.

i.e. if this is successful then TP12 should show a TTL logic 1 (>2.5V but typically about +3.5 to +4V DC)

If the counter fails this test then you will get a logic 0 here. Typically less than 0.5V DC.

It's hard for me to be sure if your scope results are good because I don't know where the 0V reference level is.

The other thing is that the diodes CR1 and CR3 must be very sensitive diodes to work at these signal levels so don't remove them or overheat them with a soldering iron and don't attach clippy type test leads to them that might physically stress or crack them. They could be expensive...?

[Chris Wilson]

I'll do the TP12 test again and show the reference level. I put Q12 back after removing the 1k test resistor and with Q12 reinstated TP12 does NOT go high when 50 MHz at -20dBm is injected to the Band 2 input. It DOES go high as soon as I warm Q12, but I am not sure if that's to be expected.

I also renewed Q19 with no effects.

Replacements for U4 and U5 have arrived, is it worth replacing them at this stage?

Thanks for persevering with me.

[G0HZU_JMR]

I don't think it's the right time to mess with desoldering U4 and U5 yet.


What isn't clear to me is if the counter sits there looping through the fig 4.3 flowchart all the time if there's no input to the counter.

Can you try looking at TP4 on the scope with nothing fed to the range 2 input and see if TP4 is changing logic state from 4V to 0V and back fairly regularly?

Because of the way the comparator operates I now suspect a healthy counter may just sit there fishing across all three ranges of range 2. eg 2A, 2B and 2C if there is no input.

So the moment a signal is connected by you it could be anywhere in the fig 4.3 flowchart routine.

So maybe it gets stuck for some reason (eg a fault) and then you connect your 20MHz test signal and heat Q12 and this (illegally) forces the signal across to the IF by bypassing the MX1 mixer and the counter program reacts in some way to this illegal entry signal at the IF and it can somehow then get back to the start of the routine and then lock onto the 20MHz sgnal correctly?

But that is just a guess....

I think we need to understand if it does indeed sit there fishing across 2A, 2B and 2C in the fig 4.3 flowchart when there is no input to the counter.
Maybe it gets part way through and gets stuck for some reason?

[Chris Wilson]

I have now refitted the new Q12, does that matter for this test? See post #65. I assume I am just looking for the voltage level change and not trying to capture a waveform of rise time or anything trick? Cheers. Let me know when you have enough of all this nonsense

OK, the trace from TP4 does different things depending what band it is on.
Bands 1 and 3 it just sits there. On Band 2 it gives what appears to be two traces. Changing the timebase gives the other trace I show linked. On bands 1 and 3 only a single trace appears. So Band 2 is doing something different at TP4, with no input signal. Warm Q12 with an input signal to Band 2 and the two traces change to a single trace. Once it cools the two traces return.

Traces:


http://www.gatesgarth.com/scope/TP4-band1.jpg


http://www.gatesgarth.com/scope/TP4-band2.jpg



http://www.gatesgarth.com/scope/TP4-...t-timebase.jpg



http://www.gatesgarth.com/scope/TP4-...put-signal.jpg

[G0HZU_JMR]

OK this kind of proves it does indeed try to run the 4.3 flowchart continuously.

The only time TP4 needs to go low is on range 2A (10-185MHz input) as TP4 puts the mixer into bypass mode with the DC bias from Q12.

It looks like it is doing this but for only 400us?

If so then it is going to need to do a very quick sniff of the comparator to see if it has responded to your 20MHz-50MHz input.

Can you put the second trace on the scope onto TP12 (comparator output) at the same time?

i.e. have both traces together to see if the comparator is flagging it has detected a bigger signal at the IF detector wrt the RF detector within this timeframe.

400us is a very short detection window and TP12 has to go to logic 1 within this timeframe or I assume the counter will conclude there is no signal being injected at range 2 input below about 190MHz if the comparator doesn't report the signal quickly enough?

We may be getting close to the problem here... (famous last words)

[Chris Wilson]

OK, will do, also sent you a PM, not sure if it helps!

Just awaiting a customer collecting an engine, and paying for the Christmas turkey, hopefully. I'll try and get you the 2 traces in a while, watch this space. Thanks Jeremy!

[Chris Wilson]

http://www.gatesgarth.com/scope/both.jpg

Top trace is from TP12

Lower trace is from TP4

No signal input, set to Band 2

~~~~~~~~~~~~~~~~~~~~~

http://www.gatesgarth.com/scope/band2-184mhz-in.jpg

http://www.gatesgarth.com/scope/band2-185mhz-in.jpg

http://www.gatesgarth.com/scope/band2-50mhz-in.jpg

Not 100% sure I am capturing what you want / expect. Cheers.

[Chris Wilson]

184 MHz into Band 2 top trace TP12, bottom trace TP4. Scope set for AC

Put 185 MHz in and both traces flat.

http://www.gatesgarth.com/scope/184mhz-in-set-ac.jpg

[G0HZU_JMR]

I can't really make any sense out of your scope displays above I'm afraid. You said you are using a Tek 7854 and that is a very powerful instrument.

On your readouts I have no idea if the two traces are showing the same instance in time as they track across the screen? Are you using a delayed timebase mode?

As a confidence check have you temporarily put both probes on TP4 to make sure they show the same trace shape one above the other and then just swap one probe to TP12?

Just do the 50MHz -20dBm test at the range 2 input for now.

[Chris Wilson]

Hmm, a probe cable was knackered, I haven't had cause to use the second probe for months. I have got my basic USB scope out and have another capture. 50MHz into Band 2 input. Orange trace is TP4. Yellow trace is TP5. My sanity check showed duplicate traces, you can at least see the settings with the USB one.

http://www.gatesgarth.com/scope/capture1.jpg

[G0HZU_JMR]

OK the USB scope looks better but you need to make sure both traces are triggered from the falling edge of TP4. I assume you do this by setting both trigger clickys to channel A falling edge?

That way they both track in time the same.


Can you repeat and look at TP4 and TP12? (i.e. TP12 and not TP5)

50MHz -10dBm sine wave into range 2 input.

TP12 is the output of the comparator. I'm hoping it would show a logic 1 shortly after TP4 falls to 0V.

There isn't much in the way of delay anywhere in the signal path after the mixer MX1.

The 200MHz LPF should only have a few ns of delay and the Q8, Q9 amplifier will have even less delay.

The comparator should be really fast so the only delay will be in the detector time constant set by C38 and this will be less than 100us.

So as long as the mixer MX1 is being biased on quickly and adequately by TP4 and Q12 then (on a healthy counter) I'd expect to see TP12 shoot to a logic 1 very soon after TP4 falls to 0V each time. (assuming you are feeding in a sine wave at 50MHz to the range 2 input)

[Chris Wilson]

Sorry, that was a typo,TP5 (yellow trace) is, and was, on TP12.

Here's a new trace triggering both on TP4. I think I see what you mean by TP12 showing logic 1. If I up the frequency I think it is doing this, and showing it properly on the USB scope now.

http://www.gatesgarth.com/scope/capture2.jpg

[Chris Wilson]

Just so I know what I am looking for, is this the signal that we want to achieve, once things are working correctly, but below the 158 MHz I am feeding at -20dBm into Band 2 to get it? Are the settings of the scope correct?

http://www.gatesgarth.com/scope/capture3.jpg

[G0HZU_JMR]

This is a quick time domain chart showing how the counter validates an RF input is in the 10-190MHz range.

Only a signal that is less than about 200MHz can get through the signal path if MX1 is biased on with the DC pulse from TP4. So if the IF detector CR3 is bigger than the RF detector CR1 then the comparator should alert the counter's processor via TP12 and the PIA.

I'm not sure about how your scope is set up but ideally you should trigger both channels from channel A so they both track together in time. But it's probably OK as it is for the purposes of this quick test.

I would imagine that once the processor gets the 4V pulse from the TP12 signal then it will hold TP4 on for a lot longer than just 1ms as TP4 needs to hold the MX1 mixer in bypass mode once the processor program progresses to actually feeding the signal to the counter itself (using U4 and U5).

So presumably once the counter processor declares the signal as being a genuine range 2A signal TP4 will then be made to stay on for at least as as long as the counter gate time?

The 158MHz plot looks a bit odd to me if this is meant to be a range 2A signal. Is this a typo and really a >185MHz signal in range 2B?

[Chris Wilson]

OK, I think I follow that Jeremy. I have removed one of the triggers and am now triggering both channels off channel A, Not only was one probe lead for the Tek 7854 wonky, but the dual trace amp has also gone noisy on channel 2, probably a switch contact. So I'll stick with the USB one, for sure!

The capture in my post #76 is not a typo, it's definitely a 158 MHz input at -20 dBm into the Band 2 socket. Here's another but at -10 dBm, taken just now, same 158 MHz input. The counter display itself is all zeros.

http://www.gatesgarth.com/scope/capture4.jpg

[G0HZU_JMR]

Here is my initial interpretation of your plot above now you have confirmed it is 158MHz.


The narrow TP4 pulse down to 0V is where the counter is at the start of the 4.3 flowchart and is is fishing to see if this 158MHz signal is in the range 10-190MHz for that 1ms window.

Now we know that 158Mhz IS in this range 2A but your counter is faulty so it fails to trip the comparator at TP12.


So it then logically goes to stage 2 of the flowchart where it turns the VCO on at 370MHz and TP4 is released back up to 4V to rebalance the mixer (so it can be driven correctly by the 370MHz LO)

Now 370MHz - 158MHz = 212MHz. So the mixer will spit out 212MHz at pin 4 and this will enter the 200MHz LPF section.

This is close enough to 200MHz to sail through the 200MHz LPF after MX1 and give a big fat detector pulse at CR3.

So comparator TP12 will go high to 4V (as per your scope plot) because CR3 detV is bigger than CR1 detV. So this is what I would expect to happen here if it failed the first test due to a fault.

TP12 presumably stays high for a longer period because the counter needs to measure this for a whole gate period rather than just a little 1ms sniff.

But the counter is going to reject this as a valid range 2B signal because the counter will be reporting a count of 212MHz. The processor program will reject 212MHz and keep trying to interpret what is REALLY being injected at the range 2 input.

Not sure what it will do next but I assume it will retune the VCO up in frequency to try and make sense of the 158MHz signal.

Presumably it will NOT make sense of it as it isn't a valid range 2B signal so at some point it will retune the VCO such that the 212MHz signal at CR3 will shift high enough to get rejected by the 200MHz LPF and CR3 will lose level and so the comparator will trip TP12 back to 0V.


This appears to be what is happening in your scope plot.

You can see the counter revisiting the whole loop again and again and failing every time to realise that 158MHz is a valid range 2A signal. i.e. you get to see regular TP4 negative pulses and regular TP12 upward pulses caused by the phantom 212MHz signal.

So either the mixer MX1 is duff (unlikely) or there's something very odd about TP4 driving Q12 that makes it deaf when MX1 is DC biased.

Hope this makes sense....

PS I've got plenty of TFM-2 mixers here that have been salvaged from unused PCBs from 20yrs ago at my place of work. They were fitted to PCBs that never even got to be powered up before they were replaced by newer designs. I can send you one FOC in an envelope if you want to try one? Otherwise you could have some grief trying to buy a new one cheaply.

[Chris Wilson]

I am sure it makes perfect sense Jeremy, and I *THINK* I follow most of it. Is there any further test possible to isolate it further, even if it means using the SA (with great care), or my other (working fine) counter / timer? Thanks for your perseverance!!


Oh, and I have fixed the 7854, the probe lead was easy, but I took the coward's way out with the noisy 7A26 dual trace plug in and fitted another I had with a broken knob, but that actually seems to work properly on both channels. Swapped knobs and a painless repair!