Output Transistor blowing - B&O Beomaster 901

[Marconi_MPT4]

Looking at the schematic 2TR33 is not related to the power amplifier section, it is just a regulated supply for the tone, pre-amplifier, and other sections. If +32V is present on the collector of 2TR33, then it is working correctly.

Are your replacement TIP120/125 transistors branded? I have had a few issues with modern replacement devices not performing to spec, even from the same manufacturer. From time to time they change the fabrication process and its the subtle differences that matter.

Replacement of all five transistors is related to the respective power amp e.g. 2TR27 ~ 31 and just advisement from B&O technical. These units could be a PITA so it was essentially to save time and potential future call-backs plus having to deal with irate customers!

When the fault has been cleared, it is almost essential for the amplifier to undergo a soak test with both channels driven at half power into dummy loads. You don't want a disco inferno at the next party!

Rich

[Chris55000]

. . .A free copy of the B & O Survey Of Semiconductors can be downloaded from here :–

https://freeservicemanuals.info/en/b...uctors%2002-82

Chris Williams

PS!

. . .Give me a collection of direct–coupled transistors for a hi–fi power amplifier anytime, having just had two TDA7293s explode on me for no rhyme nor reason (tale plus circuits will be chronicled in due course!) – I can deal with separate power transistors with far more enthusiasm than these horrible power chips!

[stuarth]

Back in post 22, you reported the voltage on T30 collector as 46.7V. This is the supply voltage, nominally 48V.

So it doesn’t look like the lamp limiter is losing much voltage.

It’s a good idea to change all 5 transistors in one go in an amp like this, as suggested by B&O, especially one with repeated blowups! If you leave out one, and it’s been damaged previously, it can trigger the failure of the newly replaced transistors.

Next time you remove the transistors in the faulty channel, and before you fit the new transistors, check the voltages on the good channel without the lamp limiter in circuit. Maybe the 48V is going higher than it should.

I’d still advocate putting a series resistor of around 270R 5 Watt between the main reservoir and the amp while you do the test with the next batch of transistors. It could save another blowup. On the pcb layout you posted, it look like there are links you could remove to add the resistor, but your board may be different. You could add the resistor between the collector of the top power transistor and the 48V rail.

Before powering on next time, you might also consider shorting out the Vbe multiplier transistor to eliminate the quiescent current setting components as a cause of the blowups.

Stuart

[paul1962]

Please forgive the interjection, but I have read this thread with interest.

Going back to my train of thought and the original scenario of Lamp Limiter (bulb), nothing blows. When full mains applied, boom.

This may be an amateur's view, but I ask myself what does a lamp limiter do ? .. limit current. I accept that under ohms law (current, voltage and resistance are linked, and as such any change in one will affect another), but in my understanding, using a limiter doesn't change voltage per se and again, in my limited knowledge, any readings of voltage is possibly not the real issue, but rather current, as this appears to cause the boom factor.

Maybe over simplistic, naive, plain idiotic or irrelevent, but just my thoughts on the situation.

Kind Regards

[Radio Wrangler]

The voltage drop in the lamp limiter will reduce the supply voltages fed to the amplifier, reducing the stress (voltage) on the transistors.

A transistor can be damaged without completely destroying it. It is quite common for a device to be degraded rather than outright killed. Gain can be reduced, Ft can be reduced, and breakdown voltages can be reduced. So you have an amplifier which runs up and looks OK on reduced mains fed into it, but which blows up when met with full stresses. This is not uncommon.

You take the amplifier apart, remove all the transistors and test them with one of those convenient little hand-held testers which are all over the place nowadays. Being battery-powered with a limited supply voltage and current, it can spot fully killed transistors and (usually) identify the pins for you, but it has no hope of telling you if the thing will break down at less than its specified voltage, or if the current gain fades at rated current. The readings from these things, especially for power devices is binary. You get "That's definitely dead" or "Dunno" with no better clues.

So B&O's advice to swap the lot is sound advice. It isn't just saving the repairer's time, which for a hobby could well be free, it also saves multiple goes at replacing the same devices.

Apart from the power devices, the smaller ones ought to be relatively cheap and easier to source so no great cost in getting them. But a dead driver VAS etc can trash the new, expensive, power devices.

Experience doesn't help you take any shortcuts. Experience tells you not to even think of taking a short cut

Solid state power amps aren't too difficult to fix. They can be handled and they need not be scary. You just have to accept that what looks like the long way round is the ONLY way round.

If you're doing a job for someone else, cost a full set of semiconductors. While you have them out, it's easy to check all the resistors, pots, capacitors. If your soldering skills are good, set the pots to their safe positions, assemble and expect a goer first time. Set the quiescent current, set the DC centring and you're there.

B&O try to sell their stuff for quite a premium as high quality gear, not just for the styling. But a look at this amplifier, as with much of their electronics, shows it to be rather basic and not at all special.

David

[PaulR]

Originally Posted by Radio Wrangler

B&O try to sell their stuff for quite a premium as high quality gear, not just for the styling. But a look at this amplifier, as with much of their electronics, shows it to be rather basic and not at all special.

David

There is a hi-fi shop in Southport which had regular demonstrations in a medium sized theatre/hall in the town in the 1970s (the shop is still there). It was usually Harman Kardon stuff together with large AR speakers. I seem to remember that electrostatic Quads were also demonstrated. It always sounded very good.

A rather posh department store decided to do the same with B&O equipment which was what the mainly sold. At the interval some audience members complained that they were using the theatre's PA system to amplify the B&O equipment. After this the demonstrators apologised and said they hadn't realised that the PA was switched on. They proceeded to crank up the B&O as far as it would go to try to fill the hall without the PA. It sounded awful and ever since I have thought that B&O is a triumph of form over function. They never did it again!

[retailer]

I may be barking up the wrong tree here but I'll tell the story anyway- the first post in this thread states the outpout devices are TIP120/125 my Motorola data book lists TIP120/125 as darlington devices - I had a very similar issue with a Crate GX212 100W guitar amp which was fitted with parallel TIP142/147 darlingtons - it was ok on a lamp limiter which limits the input power but would blow the output devices once switched over to full mains and any attempt was made to set the bias - I was using SGS branded output devices - I eventually fitted higher value emitter resistors which limited the current through the outputs just so I could run it on full mains, set the bias and measure voltages - it's been over 10yrs now and my memory is a bit hazy here but I think I may have initially gone from 0.2ohms to 22ohms, with the bias set, measuring voltages showed nothing unusual, it was frustrating, then I came across a post on a guitar amp forum regarding a similar issue with a Fender amp, which was fixed once genuine Motorola output devices were fitted.

A darlington is not just a straight bipolar but does have resistors and a diode on the device, I've always considered SGS semis as half decent, but with darlingtons there is no way of telling what is inside with regard to extra resistors and diode - with nothing left to lose I purchased some NOS Fairchild and Motorola outputs, had to buy a full sleeve of each which was not great but then they were cheap at around 80 cents ea.
This did the trick, switched back to 0.2ohm emitter resistors was able to set the bias, the amp was used by my son for a number of gigs before he sold it off to a mate.

[dave cox]

Originally Posted by paul1962

Please forgive the interjection, but I have read this thread with interest.

Going back to my train of thought and the original scenario of Lamp Limiter (bulb), nothing blows. When full mains applied, boom.

This may be an amateur's view, but I ask myself what does a lamp limiter do ? .. limit current. I accept that under ohms law (current, voltage and resistance are linked, and as such any change in one will affect another), but in my understanding, using a limiter doesn't change voltage per se and again, in my limited knowledge, any readings of voltage is possibly not the real issue, but rather current, as this appears to cause the boom factor.

Maybe over simplistic, naive, plain idiotic or irrelevent, but just my thoughts on the situation.

Kind Regards

A good question, and in this specific case we can only really guess as there is not actually enough information. However, I have seen a very similar case, with a quite similar amplifier topology [1] where the amplifier 'seemed' quite happy operating with a lamp limiter / current limited PSU / resistors in series with the PSU (at full voltage) but given a full beans startup it would self destruct in an instant. IIRC it did this at least 3 times with a fresh set of tested semiconductors each time I did manage to capture it in the act (a storage scope with DEEP memory is just about the only thing for the job) and sure enough it was oscillating rail to rail as the PSU came up then something expired rather quickly! Looking at the working channel it also did this but at a lower amplitude and damped out quite quickly. More careful examination revealed that there was somewhat more gain at HF with the new silicon and it was enough to push a rather iffy startup into a catastrophic one.

I've also worked on a amp that would absolutely not startup with a 'dim bulb' but would work with the full fat supply

dc

[1] The lower leg of the output was quasi-complementary but otherwise very similar topology. Kind of simple and expected to just work !

[stuarth]

If I’ve followed the course of this amp correctly, it’s had transistors replaced, worked ok with the lamp limiter, them blew immediately on full mains, and this has been repeated more than once.

If the supply voltage with and without the lamp limiter is 46.7V and 48v respectively, it would be remarkable if such a small change in voltage hit some sort of device breakdown, or even led to catastrophic oscillations in the same tiny voltage window for more than one set of transistors.

So what is not being changed when the transistors are changed?

Maybe the supply voltage is considerably above 48V without the lamp limiter.
Maybe all 5 transistors are not being replaced each time, and there’s a wounded transistor in there.
Maybe the quiescent current setting parts around the Vbe multiplier (T29, another Darlington) are faulty.
Maybe something on the board, a metal whisker or similar is breaking down.
Maybe there’s an electrolytic somewhere (with its polarity reversed?) which is breaking down.
Maybe it’s not the absolute supply voltage, but the switch on surge without the limiter.
Anyone want to add to this list of things to check?

It’s all very intriguing. And frustrating, these things are usually fixed by a new set of transistors, and sometimes emitter resistors and quiescent current setting resistors.

Stuart

[HatOfTheCat]

Some REALLY interesting discussion here, hoping that means I haven't just done something totally stupid and this is where people are coming for fun I'm waiting on parts so testing on hold for a few days but a summary of where I am and also to say thanks again for all the replies.

I've ordered "new old stock" Texas Instruments (as per originals) of TIP120 & TIP125 from Langrex, two of each, still mulling if I do both channels at once as Left has been without issue so far and works on full mains power. If the iron is warm I may as well.

The "fifth" transistor on the heatsink TR33, a BD238 is still the original one, the "replace all five" was from Marconi_MPT4's knowledge of the B&O normal repair procedures but seemed to think that as +32v rail was steady with no problems this transistor is actually ok. I have in any case ordered a replacement one from Switch and I can recheck old one and it with the output transistors out of the board with full mains power.

Everything on the heatsink with new mica pads, a smear of thermal paste and new plastic top hat collars for the fixing screws. The old ones did appear exceptionally thin and cracked rather "too easily" on removal.

I also have in hand new 56pF caps to replace the one across base/collector of TR20 and 9.1v 500mW diodes to replace the zener D27.

The bias transistor TR29, which was failed after last time, has been replaced with a new MPSA13. TR20 a BC407B was replaced previously (even though old one tested okay) current one tested okay and was put back in (I have spares so can replace if thought worth it). TR27 a BC253B was replaced previously and again this one tested ok so left in place (again I have spares if worth changing anyway).

EVERY single resistor in the stage has been replaced before last testing except two, none were really bad but a couple were more than nudging spec limits. The 0.39 Ohm (a wirewound ceramic) on the emitter of TR30, tests in spec, the other is the the 8.2K (2R184 on schematic) tested and again bang on spec.

ALL the other capacitors here have been changed. That is all three of the 3000 uF, both the 100 uF (all checked & rechecked are correct orientation) and the 1 nF. The 0.1 uF was also checked off the board and fine so is original (a grey rectangle film type).

To confirm, the preset variable pot resistor for bias has been changed for a new 1K one (and has its new resistors also either side). The left one was also with same.

BTW the left channel has all the resistors changed too as well was all but the 56 pF as per the right at present. Also the 1K bias pot too.

About a page back I posted last test results for voltages with 40 watt dim bulb limiting (with "modern" TIP120 & TIP120), all looked ok but then it did before last blow up too. Once I've the new "old" transistors etc. in I'll do these again and then & also try to see if I can get a square waveform in and see the output of that for any signs, then also ramp up to a 60w dim bulb limit and ALSO try some of the extra resistor options on emitters/supply rail etc. (I now have 39 ohm wire wound aluminium case ones for TR30 that I can use in place of the ceramic 0.39 Ohm).

I read with amusement the comments about B&O being more about looks than premium internals or "all fur coat and no drawers". I have to say that it was was my reaction when I started looking for any speaker protection relay or similar Still the front panel pieces & switches are beautifully milled lumps of aluminium and the top is a solid teak board... okay, the sides are stupid plastic panels with some veneer on and don't start me on putting the headphone socket round the back If I can get the amp stage working then a far worse fate awaits in trying to restring the tuner control... the little plastic peg has broken off that holds the string onto the capstan of the tuning capacitor. Its like the "Krypton Factor" figuring out the process to to put it on

[dave cox]

Full marks to you for sticking with it
And you still have a long way to go to match the saga I linked previously

Personally, I have quite a pragmatic approach - the last amplifier I looked at was my own and it had blown an output MOSFET. The device is no longer made, much sought after and highly priced so a likely target for the fakers. I could probably get it to work with a different MOSFET but I have no schematic and the manufacturer had rubbed the numbers off the opamps. I still have the case and the TX plus a few more used components

dc

[stuarth]

Hold on, when B&O and people on here talk of replacing 5 transistors, it means the 5 transistors in one power amp channel (TR27 to 31), not the 5 transistors on the heat sink.

If one channel is working, I wouldn’t change its transistors, but I would change all the transistors, and the 4 diodes in the blown channel all in one go. Is that what you’ve been doing?

Stuart

[HatOfTheCat]

Originally Posted by stuarth

If one channel is working, I wouldn’t change its transistors, but I would change all the transistors, and the 4 diodes in the blown channel all in one go. Is that what you’ve been doing?

Thanks for clarifying, I was starting to get confused. Just to confirm its only the RIGHT channel transistors & diodes I have been changing as a set. Meanwhile ALL the transistors and diodes in the LEFT channel are all the originals and untouched. That's why I was baulking at idea of changing any of them on the LEFT or using, per suggested, using the known "good" original transistors from the left to test in place on the right.

[stuarth]

Ah, that’s good. I’d leave the working channel alone, it’s useful for comparing voltage and resistance values between a good and a suspect channel.

Are you going to change all 5 transistors in the faulty channel before you next power it up?

Stuart

[HatOfTheCat]

Originally Posted by stuarth

Ah, that’s good. I’d leave the working channel alone, it’s useful for comparing voltage and resistance values between a good and a suspect channel.

Are you going to change all 5 transistors in the faulty channel before you next power it up?

Stuart

Yup, I'm going to change ALL five as I've already done the TR29 bias one (an MPSA13, has only been used dim bulb limited just to do the last set of voltage measurements). TIP120 & TIP125 will be new "new old stock" Texas Instruments ones and TR28 (BC407B) & TR27 (BC253B) will be new ones out of packet. The 4 diodes have only again been used for that set of dim bulb limited voltage measurements and have checked fine. I'll also be changing the C122 (56pF) for a new one.

Left channel working is a bonus indeed as per last set of voltage measurements can see what is just "how it should be" or going further if I get any odd temp differences once right does run under full mains power.

PS I once had to fix the undercarriage indicator lights on my own aircraft whilst trying to land it as I didn't didn't have "three greens" lit up and couldn't be certain the wheels were actually down and locked properly... fixed that with a biro jammed into a switch... at least this one won't kill me... maybe

[Silicon]

Is there any possibility that some of the transistors are working close to their maximum voltage rating?

When I replaced some driver transistors in a Leak 70 I tested the Vce beyond their datasheet figures.

I believe Leak did this in the factory.

[HatOfTheCat]

Originally Posted by Silicon

Is there any possibility that some of the transistors are working close to their maximum voltage rating?

I did a full set of voltage readings (see reply #22 of this thread) and although taken with dim bulb limiter, none of the five transistors in the problem RIGHT channel were presenting with a Vce close to their datasheet max figures (or would be by the schematic specs).

[Philips210]

Regarding replacing the five transistors. I'd be inclined to try and obtain genuine new old stock parts. There are a few good suppliers on eBay that stock NOS components, Langrex for example which you can trust.
By fitting period transistors there's a greater chance their parameters are to be expected and the circuit will perform as intended. Some current new stock transistors can often have a higher transition frequency fT which can lead to instability.

Just another thought. You could make a series of resistance measurements throughout the amp comparing the results in each channel but watch out for misleading readings due to charging electrolytics.

Regards,
Symon

[Radio Wrangler]

So when you see those treacly words "The Sirius Cybernetics Semiconductor Company has a policy of continuous process improvement and device performance may be changed without notice", you'll understand it as the threat that it is

Not as scary as only two greens, though.

Congratulations on the fix

David

[HatOfTheCat]

UPDATE:

Here are latest "repairs" and test results for the RIGHT channel that has been the problem. Feedback on the results and votes for "next step" please... I have a pair of 47 Ohm wirewound aluminium case resistors to substitute for the two 0.39 Ohm ones on TR30 and TR31 but I also have a 100w bulb in cupboard... but not any kind of bench supply, variac or similar.

All 5 of the transistors on the RIGHT channel have been replaced:
TR30 a TIP120 "new old stock" Texas Instruments branded (as were/are the original as fitted) from Langrex
TR31 a TIP125 "new old stock" Texas Instruments branded (as were/are the original as fitted) from Langrex
TR29 a MPSA13 (the bias transistor which is thermally coupled to TR30 with blob of thermal paste as per original) from Finedon Electronics
TR28 a BC407B "new old stock" from Selectron (UK) Ltd
TR27 a BC253B from Telec Electronics in Netherlands

Also new are:
56pF cap across base/collector of TR28
9.1v zener diode D27 on the 32v rail

The FOUR diodes between TR30 and TR31 all tested perfectly

I powered up the receiver with 40w "dim bulb" without issue, checked all voltages around all the transistors and then went for it and... upped to a 60w (I'm not that brave/stupid to go for broke again )

*** ALL the below testing done while operating with 60w dim bulb as limiter ***

Attached are schematic for reference and also scope pics from tests with 1 kHz square wave input signal input the back wiring of the TAPE source panel plug. BOTH signal source and the scope are a Miniware DS211.

The four scope images are:
- raw input signal measured at source
- measured at emitter of TR30, receiver volume at zero
- measured at emitter of TR30, receiver volume at 10 (out of 12 max)
- measured at emitter of TR22 so equivalent on the LEFT channel, receiver volume at 10 (out of 12 max)

In addition 5 seconds of lively scope trace of RIGHT channel with some music pumping through, guess that tune... there are no prizes I'm cheap https://youtu.be/7L2icVX5G0g

Distortion seen on scope trace came in above volume setting of 9 and quickly built to that seen at setting of 10. First traces of distortion on BOTH channels could be seen at a little over 6 but very very slight, then a very slow increase but then falls off the cliff after 9. I couldn't really see any real difference in this between the RIGHT and LEFT channels.

Temperature tests with a thermocouple taken after switch on and running at idle for over 20 mins (so before any other tests). Measurements made with the probe pushed on the back plate of each transistor in same location, in case or variation across the package etc. Measurements were made 5 or more times for each to avoid any errors in degree of physical contact and were consistent each time. Are to nearest 1c as that's all the temperature mode on my multimeter can do. Some variation could in part be due to TR22 being at end of the heatsink whereas TR30 is in middle with transistors heating to both sides of it ? But its still a measurable difference.
TR30 30c (versus TR22 28c on LEFT)
TR31 29c (versus TR23 29c on LEFT)

I also repeated the full set of voltage measurements around each transistor with nil input signal and volume at zero. Again values in square brackets are spec from schematic and in curved brackets are as measured at equivalent points on the "good" LEFT channel:

TR27
E 17.8v [19.6v] (17.6v)
B 17.2v [19.0v] (17.0v)
C 0.6v [0.6v] (0.6v)

TR28
E 0v [GND] (0v)
B 0.6v [0.6v] (0.6v)
C 20.0v [21.0v] (19.9v)

TR29
E 20.2v [22.5v] (20.1v)
B 21.3v [?] (21.3v)
C 22.6v [25v] (22.5v)

TR30
E 21.4v [23v] (21.3v)
B 22.6v [25v] (22.5v)
C 45.3v [48v] (45.3)

TR31
E 21.3v [23v] (21.4v)
B 20.2v [22.5v] (20.2v)
C 0v [GND] (0v)