Tek 2225 Noisy PSU

[TaylorD93]

Hi Folks,

Im hoping the forum can help me with a repair project ive been struggling with, which i cannot seem to solve. I have a Tektronix 2225 50MHz BW CRT Scope. Just like the one which featured in one of Daves videos. It has two issues which i believe are linked, but am not completely sure. There are two conditions i have noticed;

1) With no probes connected, with AC or DC coupling on a selected input (say CH1) on 5mV/div with the x10 Y (giving 500uV/div resolution) i am measuring 18KHz noise. Does it on both CH1 and CH2, is in perfect phase with the PSU switching cct too.

2) With the Input coupling to GND, on 50nS/Div (i.e the fastest timebase) to 0.2uS/div there is distortion at the start of the CRT beam. With the input grounded, the amplitude of the "squiggle" stays the same on all Volt/Division scales, constant with X or Y adjustments. It is "stretched" when the X axis is set to x5, x10 etc.


I had actually noticed symptom 2 first (the squiggle). Given it was constant no matter the voltage division, i didnt think it was between the Input or the attenuator, so started at the Output of the Final Y amp and traced back from the Y Final amp. I found the attached waveform to the input of the Y Final amp stage from the delay line as well as the input of the "Delay line driver circuit". I didnt bother to trace it back further, because when i started probing on the +8.6/-8.6V rails, i found the noise was there too. This squiggle is visible on all input signals, such as a 20MHz sinewave input, you can see this signal superimposed onto the Sinewave, this is both visible on the CRT and when probing the Y input with my other scope. When changing time base to say 0.1uS or 0.2uS, you can see the squiggle get condensed, which says to me it could be occurring at the same rate as the sweep rate of the scope.

I assumed that this noise was coming from the PSU, this was supported by a few online posts that these Scopes do suffer from Switch mode noise so a common approach is to "replace all the electrolytic caps" so i did just that. All Caps within the PSU Area of the board, were replaced with the best quality caps i can get, the lowest ESR etc.

After this i measured and adjusted the -8.6V rail as per the service manual,

-8.6 = -8.6103
+5.1 = 5.1087
+8.6 = 8.7646
+38 = 37.606
+100 = +97.917

Now while all of these DC readings are within their respective limits. The ripple/Vpk to pk readings are not so. The +100 and +38V rails are in spec, but the +/-8.6 and +5.1V are all out of limits on the noise. I did my best to measure the noise with a x1 Probe (as per the service manual) and a short scope spring for the reference (avoiding the croc clip on the probe). The "typical" noise is in spec, but there are higher voltage spikes occurring on the rail

After changing the PSU Caps, the noise on the display was no better either.

The PSU inverter, which is based on a TL594, and drives a push pull inverter using a number of BJT devices, I probed the bases of Q930/Q960 and while the Service manual shows a trapezoidal waveform (imagine a triangle wave with flat spots on the peak/trough), mine was "skewed" and had some ringing on the rising and falling edges (will try and get a scope capture for a follow up). However the ringing didnt quite match up with the "squiggles" on the display. However, the noise measured on the 500uV/Div resolution (symptom No.1), was perfectly triggered to that of the Inverter switching.

Incase it was switching noise from a low line voltage, i did put the scope on my Variac and bumped it up to just over 240VAC (was a typical 230VAC during previous testing) and the noise was the same with no changes at all.


Im a stumped as to where to really go next, the main bulk caps are now replaced, while there are some 15uF caps dotted around the board which i have not replaced, I cannot actually find anything wrong with the scope, but it has a seemingly very noisy PSU.


As the 2nd symptom appears to be in sync with the Sweep rate, ill do some digging around that circuit, and get traces of the noise i am measuring on the PSU rails and will feed back any progress.



Id really appreciate any help with this, marginal or noise issues like this are my biggest weakness when fault finding, ive spent months on/off with this and i really wish i could solve it.

Thanks,
Taylor

[floobydust]

Commonly it's caused by the electrolytic capacitors failing in the power supply.
You get high ripple voltage which can heterodyne with the scope's sweep speed and cause strange effects.
Instead of troubleshooting them, just outright replace them ALL in the power supply secondary at least. I put a 2235/2235b re-cap spreadsheet here.
The Sangamo caps crack and leak, others dry out.

[TaylorD93]

Commonly it's caused by the electrolytic capacitors failing in the power supply.
You get high ripple voltage which can heterodyne with the scope's sweep speed and cause strange effects.
Instead of troubleshooting them, just outright replace them ALL in the power supply secondary at least. I put a 2235/2235b re-cap spreadsheet here.
The Sangamo caps crack and leak, others dry out.

In the PSU area, i have replaced them all, from the 2200uF 80V at the main DC bulk capacitance, through to the 1000uF caps on the +/-8.6 rails.

I will see if i have a close equivalent to the 15uF caps on the rest of the board, i think there is also a 100uF underneath the Y attenuator board but access will be a bit tight.

[bdunham7]

I had a 2213A with a ~20kHz noise that would just not go away and it turned out to be some issue with the transformer itself.  I don't know what the fault was, but it was radiating EM energy at 20kHz so much that with the cover off of the PSU I could pick it up with an oscilloscope probe just hanging in the air 3 feet away.  Also, make sure you have the PSU and shields fully assembled when testing, at least to the extent possible.  Even properly functioning PSUs radiate a bit of EM in these.

[wn1fju]

For what it's worth....   I had a Tektronix 2215A that was exhibiting ~25 kHz noise on the 5 mv/div and 2 mv/div scales with no input attached.  It seemed to be more noticeable on channel #1 than on channel #2.  I also suspected something amiss with the switching power supply.  But then I had the scope open on the bench and gently twisted the metal frame a bit.  The noise magically went away.  I traced the fault down to a little metal ground clip at the back of the cylindrical metal shield covering the CRT.  Cleaned up the corrosion on the clip and shield, reinstalled the clip, and never had another noise problem again.  Go figure!  So, check all the mechanical ground integrities.

[David Hess]

2) With the Input coupling to GND, on 50nS/Div (i.e the fastest timebase) to 0.2uS/div there is distortion at the start of the CRT beam. With the input grounded, the amplitude of the "squiggle" stays the same on all Volt/Division scales, constant with X or Y adjustments. It is "stretched" when the X axis is set to x5, x10 etc.

There is usually some ripple in the trace from the gate signal which controls the sweep.

I assumed that this noise was coming from the PSU, this was supported by a few online posts that these Scopes do suffer from Switch mode noise so a common approach is to "replace all the electrolytic caps" so i did just that. All Caps within the PSU Area of the board, were replaced with the best quality caps i can get, the lowest ESR etc.

...

Now while all of these DC readings are within their respective limits. The ripple/Vpk to pk readings are not so. The +100 and +38V rails are in spec, but the +/-8.6 and +5.1V are all out of limits on the noise. I did my best to measure the noise with a x1 Probe (as per the service manual) and a short scope spring for the reference (avoiding the croc clip on the probe). The "typical" noise is in spec, but there are higher voltage spikes occurring on the rail

After changing the PSU Caps, the noise on the display was no better either.

Make sure that the noise you are measuring is not an artifact of the measurement.  When you connect the probe, first connect the probe tip to the same spot as the ground clip and see what noise is found.

You might need to make a differential measurement to get an accurate measurement.

[TaylorD93]

...

Make sure that the noise you are measuring is not an artifact of the measurement.  When you connect the probe, first connect the probe tip to the same spot as the ground clip and see what noise is found.

You might need to make a differential measurement to get an accurate measurement.

I was probing directly across the main bulk caps of each of the supply rails, the ground croc clip wasn't used, I have the spring probe which slides over the tip of the scope probe to keep the loop as small as possible.

I'll try a differential measurement when I'm home tonight.


I did find that sitting my probe in the vicinity of the psu, it detects the greatest radiated emi from the HV transformer which has a tap/winding from the main switching transformer.

[TaylorD93]

So I have measured the following, all with the probe on x1 and the Rigol scope set with the full bandwidth etc. It was measured directly across the decoupling cap on each supply rail. I cant recall the numbers off the top of my head.

I put the scope back inside its enclosure and tightened it all up (after checking all the chassis connections) the Noise measured on Ch1/Ch2 on the 500uV/Div resolution, dropped down to 250uV peak to peak, surely that still isnt correct? its half of the reading outside the case.

[TaylorD93]

These traces are on 50nS/Div and are on the X amp circuits. they were measured at U745/U755 which are multiple transistors in a single package, on the attenuator board, there are some very high freq spikes which appear to occur when the ramp starts/stops. The amplitude of this spike doesnt change when changing the time base, but at the high frequency sweeps, it does make a difference.

you can see here that when i zoomed in on the U745C Pin 6 rising edge, there is some overshoot which seems to match the ringing i am seeing on the output of this circuit down to the A1 board.

U755C Pin 8 for example, is sinking current from the X amplifier stage down on the main board, but i cannot see which "end" of the cable (from the attenuator board to the main board) is causing this High Freq switching spike, is it at the Final Amp stage, or local to the Attenuator board?

[TaylorD93]

Finally without spamming the forum with my scope captures, if i put the scope in XY mode, the 4 wire ribbon which goes from the Attenuator board, down to the Main board, which carries the Sweep signal, has GND, X+, X-, GND (Pin 1 - 4) the PCB tracking is then routed along the edge of the PSU circuit.

As XY mode is essentially a stationary sweep, i have probed on the X+ and X- signals, i have an exact match for the noise coupled on the +100V Rail! exact same phase and waveform shape.

So is it that the noise is coupled from the nearby PSU circuit to those traces? or from the +100V feed into the X Final amplifier stage? Im not sure how to tell the difference.

Edit: So the 100V rail goes via a Wire Link W985, to the 3 pull up resistors on the X Amp output stage, with the W985 link removed, there is no longer any noise on the X+ and X-, likewise the noise on the 100V rail is also significantly reduced, although it is still showing an 18KHz frequency, that may be the nature of the beast so to speak. There isnt much the 100V rail is connected to, could it be old/aged output transistors?

[TaylorD93]

Im really struggling to find out the cause of these issues.

C914 which is part of the smoothing of the first "pre regulator" typically 15uF, was out of spec so i had changed it to a 47uF (Closest value i had with sufficient voltage rating) but that resulted in no change to the performance one of the chokes started audibly ringing, i suspect the change in capacitance made the frequency shift to that closer to the physical resonance of the inductor and started to sing.

Im stlll none the wiser either to the distortion on the Y axis which occurs when the final stage is powered from the 100V rail.

As suggested in this thread i have checked all the chassis ground connections, cracking all screws loose and then retightening them, i have good chassis grounding through the scope.

[floobydust]

I would confirm your Rigol is not adding garbage to your waveforms. You could easily be going on a goose chase looking at 50nsec blips the scope is making up. I've seen them do it and don't take them seriously. Sometimes it's literally comedy what they display in the midst of SMPS EMI.
Is there another hypothesis for the 2225's problem you have? I would roll back and reconsider a different approach.