Uher CR240 Troubleshooting Best Practice

[Uncle Bulgaria]

Bear with me here as I'm trying to work through this logically, but it may seem far from logical to those in the know!

From the reading I've done on the BJT amplifier topologies (that Wikipedia thing where the first few sections make sense, then the advanced mathematics lose me entirely) this is a common emitter circuit of three stages. The signal is cascaded from the base of T301 to the base of T303 and subsequently T305, with the voltage taken from the collectors.

The supply is the 6.2V at pin 315, and all three transistors for the left channel are biased through emitter resistors bypassed by capacitors. The supply is common to left and right so issues here would affect both channels.

If the transistors are not to blame for the failing signal, and I have checked the resistors, then the only things left are the capacitors between the terminals of the transistors.

Assumptions:
The capacitor between collector and base is for mitigating the Miller effect.
The fact that the signal fades predictably and creates the same flattened waveform, before rising on switch-off, suggests a capacitor charging/discharging and a filter being applied.

Should I therefore be looking at C312 & C313? If C312 is a Miller capacitor, what is C313 for? In the explanatory topology diagrams I have found, these capacitors are not included and I don't understand why there is one between the base and emitter. If for blocking, why not just not include it so the emitter and base are not connected?

[ms660]

Don't get in too deep...

Common emitter T301, T303...input to base output taken from the collector, common collector T305 (aka emitter follower)...input to base output taken from the emitter.

Common base...input to emitter output taken from the collector, no common base amp in your unit so far as I can see.

C312 shunts some of the HF to ground, remember that on playback the HF signal coming from the head will be greater than the LF signals.

I would check the transistor voltages under fault conditions then take it from there, I don't know what they should be off hand because I don't have the full manual but the voltage difference between emitter and base should be approx. 0.6 volts +ve or thereabouts WRT the emitter if the base emitter junction is ok.

Lawrence.

[Uncle Bulgaria]

Originally Posted by ms660

Don't get in too deep...

I'm just doggy-paddling around in the shallows like a grom.

Common emitter T301, T303...input to base output taken from the collector, common collector T305 (aka emitter follower)...input to base output taken from the emitter.

Got it - I should have put the 'subsequently T305' in parentheses as it is not part of the common emitter.

C312 shunts some of the HF to ground, remember that on playback the HF signal coming from the head will be greater than the LF signals.

This is because of the equalisation, as in the RIAA curve?

...the voltage difference between emitter and base should be approx. 0.6 volts +ve or thereabouts WRT the emitter if the base emitter junction is ok.

I think this can be ruled out as I've swapped the transistors from the working channel to the failing channel with the problem still evident.

I've made some jumpers for the necessary pins and can now look at the board as it's running outside the box. I've measured the B-E voltage having verified the distorted wave is present on the collector of T303.

L channel: T301 553mV, T303 570mV, T305 600mV.
R channel: T302 553mV, T304 553mV, T306 629mV.

However, interestingly there's an anomaly with the emitter-ground voltages. The bad channel T301 is 1.1V, but its equivalent T302 is 500mV. T303 is 1.7V, but T304 is 1.3V.

I think I'm investigating the biasing...

[Uncle Bulgaria]

Huge progress!

After much metering and head scratching, I replaced C316 with an electrolytic I had - there's a sine wave on the L-channel output almost as big as the one on the right!

Great strides. If that tantalum's gone, I'm wondering about replacing the paralleled C106/107 and C110/111. The reason being that the 'scope output for both channels is very jumpy. This suggests to me a DC imbalance which could well be the biasing, based on the earlier C316 problem. Tantalums aren't cheap though, so perhaps it's something else.

Both collectors in the first stage show a jumpy wave with distorted troughs (picture), compared to the completely vertically steady input. On the collectors of T301 & T302, the peaks stay pretty still but the troughs rise and fall together, reducing the overall amplitude. Not by much (say 5-10mA from a 50mA overall) but presumably it should be as steady as the input?

[Uncle Bulgaria]

I'm really digging this oscilloscope thing now and I'm sticking its nose in everywhere.

I've put a square wave in as shown (not very clean, but obvious) and got this bizarre figure out. Does this show there's something horribly wrong with the overall amplifier's performance?

Is it time to start looking for culprits elsewhere in the larger circuit, or is the board to blame? I can't see how this would be rectified later on unless gigantic feedback is applied, and with the elucidation of Lawrence this appears to be quite a straightforward set of amplification stages with the only external connection being to the equalisation switch or subsequent Dolby.

Operating the equalisation switch between the different tape types displays visual changes on the 'scope, but they're minor compared to the overall shape of the waveform.

[Ted Kendall]

Looks like you're overloading something there. Try reducing the input amplitude.

[Uncle Bulgaria]

I gave that a go, thanks Ted - however, it just evenly reduces the amplitude of the odd shape. Interestingly, attached is a picture of the output from a sine wave of the same frequency and original amplitude. It seems all right to me, so what's the square wave difference?

[Station X]

What frequency is your square wave?

[Uncle Bulgaria]

1kHz, Graham. I chose that because the 1kHz sine wave appears to be de rigeur for tracing signals, and seemed reasonable for an audio frequency amplifier.

[Station X]

When displaying the square wave, was the SG connected to the amp? ie were you measuring at the amp's input socket.

[Uncle Bulgaria]

The SG (signal generator?) is a tone app on my 'phone. I have the relevant board from the deck out on the desk so I can access the component side, with jumpers connecting the important pins to the motherboard.

The input is from the 'phone's 3.5mm TRS to three crocodile clips connected directly to the board connections they feed.

The initial trace photo above was taken at the point the input first joins the board, with the distorted wave trace taken from the output, which in the complete circuit goes to the Dolby board.

The undistorted sine wave input/output pics were of traces at the same points, with the same amplitude and frequency. The only change was to 'rectangle' from 'sine' in the drop-down menu on the app. I notice the square was much noisier than the sine on input, but not as distorted in shape as the output.

[ms660]

What's the input voltage (RMS) that you are feeding in?

It should be low, 2 to 3 mV (approx. guess) (5.5 to 8.5 mV peak to peak) I would stick to sine wave for fault finding with waveforms.

If the "good" channel is working ok use the voltages and waveforms in that channel for reference.

A voltage table would be handy....E, B & C WRT ground for the three transistors in the good channel, ditto for the faulty channel, both channels under no signal conditions.

When showing waveforms say what frequency, what point they are taken at and the actual peak to peak voltage at that point, by actual I mean the voltage at the point of measuring...eg: make sure your scope's Y amp is set to cal. and you've taken into account any setting of the 'scope's probe, eg, X1, X10 etc.

Lawrence.

[Uncle Bulgaria]

Timely advice, Lawrence. Time to correlate the findings.

Current status: changing C316 for a temporary electrolytic brought the left channel back, and it no longer drops out. With the board jumpered out on the desk, I can play a tape through headphones with both channels working. Lots of hum, but the jumper leads are very long and it could probably be explained by interference. Signal injection is made on the board so not affected by input jumper length.

Now it's the previously 'good' right channel that is evincing the peaks clipping (I know the term, but haven't seen it personally, not having used a 'scope - it looks like the peaks are bouncing up and being squashed against a hard surface). Both channels (though more the right) have noisy troughs, and both output traces are 'jittery' in that the entire trace jumps up and down the screen irregularly and many times a second, despite the input trace being steady.

Under no signal conditions, no input or output connected, the voltage to ground is shown in the attached table.

--------------------------------------------------------------------------------------------------------

Previous traces have been made from a 7mV RMS (approx.), 1kHz input. All with the probe on x1, changing the Volts/cm dial after calibrating from the 1v point on the front.

For T301, for example, this input results in 56mV output sine at C, with ~5mV noise on the trough. At E there's 1.7V (DC coupled).

The equivalent at T302 is 80-140mV on C, with the 60mV difference being the troughs jumping about in a jittery fashion while the peaks stay pretty much stationary. There's 0.5VDC on E in this case.

[ms660]

If its working ok I would leave things as is, the bouncing might be a 'scope triggering issue, don't forget that the amplifier shouldn't be overloaded, if it is the output signal will be distorted.

Lawrence.

[Uncle Bulgaria]

Originally Posted by ms660

If its working ok I would leave things as is

So you think the differing voltages under no-signal conditions are not a cause for concern?

I made the voltage measurements just now, but the signal traces and voltages last night. On trying them again now, the noise is far less so it could be environmental. Bizarre.

However, I tried reducing the signal still further, which makes the left channel go from a regular wave to bouncing and clipping, and the right from a steady wave with a pointed trough to bouncing and clipping.

Videos (Dropbox Links):

4mVpk-pk Input 879KB
Lout - 1Khz - 4Mvpk-Pk Input 1.44MB
Rout - 1Khz - 4Mvpk-Pk Input 1.44MB


A feedback issue? On the left channel there are new aluminium electrolytics in the bias section, except for the equalisation on the right of the diagram.

I can't put the board back in yet because the electrolytics are too big. From the replies on my tantalum conundrum thread, it seems the thing to do is replace the tantalums. As C316 was the original culprit here, and the testing of the removed tantalums is going to be inconclusive, I'm tempted to redo this board with small aluminium electrolytics.

[ms660]

Whoops missed the voltages, 6.2 volt rail is down you need to look into that.

EDIT: Post crossed.

Lawrence.

[Uncle Bulgaria]

Yes, that seemed odd. It seems to have fixed itself though, so I'm happy to blame human error for those readings.

Now I have a decent 6.3V on pin 315, and the collectors of T305 and T306. The others are not so good, with T303 seeing 4.1V but T304 seeing 4.9V. T301 is 2.5V, but T302 is 1.7V. As the clipping is visible on only one extreme of the wave when I boost the input, I guess this is the same as 'swing' in valves, and the biasing means the quiescent point is not in the centre of the transistor's range.

So I should be looking at the biasing rather than the collector resistors, as the bias resistors (or leaky capacitors) drawing more current will cause a larger voltage drop over the collector resistors.

A lot of this thread might appear nonsense, but even the act of writing it down here is a really helpful aid to thinking!

[Uncle Bulgaria]

It feels like the home straight now.

With the newly electrolytic'd playback amplifier, and many tantalums changed on the main board (and in the power supply), I've played a tape through headphones with both channels loud and clear. The changes have removed a lot of the hiss and hum that was present even when the channel was 'working', so I can only put it down to the new capacitors.

I've just put the meter panel back, and found a duff resistor - the panel illumination was on whether the lighting switch was on or not. Now it's fixed, but the new supposedly 3V bulbs (grain of rice/wheat types) I got are so dim the filaments are barely glowing. They measure about 2R each, compared to the 12R on the one remaining original. Is this definitive that I've been sold the wrong bulbs, so I can complain to the seller?

The circuit specifies 3V, 30mA, which seems impossible to find. I got these.

[ms660]

As a rough guess from your resistance measurements I would say the replacement bulbs are 0.15 to 0.2 Amp (ish) bulbs, that's using 8 as a multiplier from cold to hot...I=V/(8*R)

So there might be a clue there (think current)….Can't discuss further on your purchase due to forum rules about Ebay.

Lawrence.

[Uncle Bulgaria]

That's what I thought Lawrence - something not right, though I didn't know about the *8. They were only dropping about a volt each. I might replace them with LEDs.

Unfortunately something's now gone in the PSU, leaving 4V on both the 9.5V and 6.2V rails, and a smoking TBA 820.

One step forward...