A Very High-End Crow Bar

[PE9ZZ]

I don't get it. I got this very unusual, very spartan high end hybrid tube amp in with the complaint: blows its fuses. The patient is a Croft Acoustics Phono Integrated amplifier. It got some rave reviews across the internet and of course it is reassuringly expensive. My amp was bought second hand and promptly blew its fuse when the buyer hooked it up at his home after a successful demo.

I've requested the service manual/schematic but I have yet to hear back from the maker. This amp is unusual in that it is point-to-point wired using terminal strips as anchor points. It is also unusual in that the gain stage is a single half ECC83 followed by a 2SJ162/2SK1058 MOSFET pair as buffer with another MOSFET thrown in for good measure. Refer to the attached reverse engineered schematic.

Now I see two serious design flaws. The first one is causing the fuses to blow. There is a crow bar across the speaker terminals in the form of a triac. I think the idea is that when DC appears at the output it gets triggered blowing the fuses. Now look closely at the capacitor and resistor connected to its gate. Shouldn't they be swapped? This is what happened: I had a DBT hooked up, time delay expired, DC offset got via the big 2200 µF cap into the gate, boom. My DBT prevented the fuse being blown again.

The second flaw is really puzzling me. A tube amp gain stage needs a really, really quiet power supply. This one isn't. There is almost a volt AC ripple riding on the supply voltage. What is this circuit supposed to do? It certainly isn't getting rid of the ripple. Nor is it doing any stabilizing.

So my idea is to swap the crow bar components and fix the power supply by adding a 200 V zener to ground and ditching the LM317. Is this a sound idea or am I missing something?

 

[PE9ZZ]

This to illustrate the issues. This is the crow bar:


and this the weird high voltage power supply:


The LM317 is wired as a current source, sort of, but that doesn't make any sense!

 

[gadget73]

If the mosfet and the LM317 were reversed I'd say they were using the LM317 as a reference and the mosfet as the pass element, but it appears to be backward, using the mosfet as a voltage drop and the LM317 as the output. I don't understand what the purpose of that is. Seems you wouldn't get the ripple or voltage regulation benefits of using one.


triac on the output also seems strange. Most protect circuits I know of use a relay to isolate the speakers, not a triac to short it to ground. That also appears to only function as a peak detector/limiter anyway. It wouldn't do anything for DC on the output, it should only trigger when the voltage at the junction of that cap+resistor to ground exceeds the turn-on voltage for the triac.

 

[PE9ZZ]

Trigger voltage for a triac is around 0.5 V. This particular device is spec'd at 0.25 V minimum. So when even a whiff of DC appears at the output, poof. I can't remember seeing this before. Usually it's a relay.

I reversed the resistor and capacitor. No joy. It is way to sensitive. DBT saves my fuses. Maybe add another resistor to ground.

As for the power supply, the voltage goes all over the place. It sucks. I'm going to use a TL431 with a transistor as totem pole load driving the MOSFET. Rock solid 200 V.

 

[gadget73]

wouldn't the capacitor block any DC though? I just don't see how its supposed to work but I don't claim to fully understand solid state electronics by any stretch.

 

[Tom Bavis]

Which fuse blows? I think one of the output FETs is blown - a good one will measure like a diode from drain to source.

 

[PE9ZZ]

The fuse in one of the supply lines. In this case the negative. Nothing else was blown. BTW- every MOSFET already has a diode between its drain and source. This is the so-called body diode.

 

[Bill Ferris]

FWIW, my Mc. MA 6100 uses a "properly designed crowbar" DC output protection circuit, being its output circuit design is direct coupled with no auto former protection.

Maybe its circuit could be built & installed in place of the one currently in the amp in question..

A/K says my MA 6100 service manual`s pdf is too large for me to attach, so it will have to be found elsewhere, like perhaps, the A/K digital documents section.

Good luck.

Ciao

 

[gadget73]

just out of curiosity, does it stop blowing fuses if the crowbar is removed? What do you get for voltage at the speaker terminals if it will power up that way?

 

[PE9ZZ]

I have it on DBT so fuses don't blow. With the triac in place it clips at 2 Vpp. Bulb glows full on. With the T1 lead disconnected it will not clip until supply voltage (50 Vpp) and bulb stays dim. The clipping is consistent with a triac being triggered with its rated current (10 mA or so). Looks like the electrolytics (which are now across ground and gate) don't remove any AC. When I put it in LTSpice it works like a charm. I don't get it.

Edit: Oh, now I geddit! I reversed the T1 and T2 electrodes of the triac! In LTSpice it now triggers at 25 VDC. Bit high but at least it will protect the speaker by blowing the fuse. Guess you'll have to be careful lowering the needle into the groove and turn the volume down.

Leaves the power supply. The TL431 replacement I concocted works really well in LTSpice and I expect it to be dead silent when I put it in.

 

[PE9ZZ]

Leaves the power supply. The TL431 replacement I concocted works really well in LTSpice and I expect it to be dead silent when I put it in.

That was a mistake. It was not silent at all! In hindsight I think the zener diode connected to the base of the MPSA42 was causing the very audible noise:



I had to add an extra RC filter to make it silent:



Also in hindsight, the original PSU was not broken after all. It was just set at a way too high voltage:



The MOSFET has a 10 V voltage drop to ensure the LM317 won't blow up. This is much more than the raw DC can provide so it duly dumped the ripple onto the anode voltage. The replacement PSU is lower and has ample headroom to remove any ripple. Drawback is the lower headroom in the triode amplifier causing a somewhat higher distortion.

Left the blown fuse. I discovered that sometimes the poweron delay doesn't get triggered. This causes the output to be connected to the rapidly rising anode voltage of the triode. This trips the triac. I tried a lot to fix it but it would not go away:



Eventually I removed the (luckily!) socketed NE555 and replaced it with another. The old one was TI, the new one vintage Samsung. So far no complaints. I did get the LED brightness up significantly though by removing a useless resistor (R5).

Finally. Had I had the schematic right away it would have saved me precious time. But reverse engineering it was a fun exercise. Even with the triac red herring.

 

[PE9ZZ]

All this work for nought... The amp came back a few days later with a blown MPSA42:



Turns out that I overlooked the dissipation of this poor transistor. It was 2 W. It took out the zener, the TL431 and the series pass transistor (P5NK50Z). Because I did not check the pass transistor I had to replace everything again. Twice. Also learned once again that critical measurements should not be done while on DBT.

In any case, the results are excellent. The amp is dead silent as it should be.

The revised schematics: