Wednesday, February 01, 2006

Quadrevision Live: Peter's 303 Explosion

Peter wrote:

"Helaas is mijn 303 gisteren met veel brom en audio-explosies (de woofers lagen zowat in de kamer) overleden. Het bleek dat één van de voedingselco's was bezweken. Ik heb inmiddels de twee voedingselco's van 2000 mu vervangendoor één van 4700 mu en de doorgebrande zekering vervangen. Bij inschakelen ontstond een lichtflits op een versterkerbordje en toen ben ik maar gestopt...."

Unfortunately my 303 has deceded yesterday with a lot of rumble and audio-explosions (the woofers were almost laying in the middle of the room). Obviously one of the power-supply capacitors was broke. I have replaced both capacitors by one 4700 mF and I replaced the fuse. When switching it on there was a flash of light on one of the amplifier-circuits and so I decided to give it up...

This morning Peter's 303 arrived in Antwerp. A first visual inspection of the 303 does not show a nice picture, but we've seen worse than this, haven't we?

Revision Step 1: Checking the power-supply

The first step is to disconnect the power-supply from the rest of the amp and to measure the voltage. Wow !!! It shows tho be 83 Volt instead of 67 Volt as it should be.

In fact 83 Volt is the unstabilised voltage on the capacitor so it is clear that the stabiliser-circuit no longer does it's job. This has probably blown up the output-circuits as well ;-(

Removing the stabiliser-circuit and connecting it to an external stabilised power-supply soon showed that the faulty component was the small trimmer-potentiometer on the board.

It is always fascinating to find out what components in old vintage Hifi cause the problem... Usually it are the capacitors and things like potentiometers that are aging in a very bad way.

The Printed Circuit-board itself shows some burn-spots - it's had it too hot lately. But for the rest this won't influence the good functioning as far as all the components are OK.

After replacing the trimmer with a new one and calibrating it, we measure 67 Volt. Tomorrow we'll check the amplifier circuits.

Revision Step 2: Checking the amplifier-boards

It's even worse than I tought.

On the PCB-board a complete track has gone up in smoke...

This is not difficult to fix with some wire but this means there has been some very serious overload on the circuit.

The other channel still works as it should, obviously (there is 37,5 Volt on the output-line before the output capacitor).

Checking the transistors shows that, indeed, TR1, TR2, TR105 and TR107 are blown up.

TR1 & TR2 Wont' be a problem, they can be replaced by a (good quality) 2N3055, which is the most common power-transistor ever made. TR107 won't cause any problems neither, this is a simple BC109.

But what about TR105 (the PNP driver)?

In the original design "38495" (NPN) and "38496" (PNP) were used as drivers. Of course they don't exist anymore.

Some surfing on Google learned that they were replaced by BC441 and BC461 resp. Actually BC441 has been replaced by 2N5320.

So I ordered 8 of each. It's not that I like to blow them up but I want some reserve if it works;-)

And BTW if I change them in one channel I'll have to change them in the other channel too, we want exactly the same sound in both channels of a 303, don't we?

Spending my time in a good way, waiting for the transistors to be delivered, I already replaced the output-capacitors with 4700 mF 100V.

Revision Step 3: Solving logistic problems ;-(

The new BC461 that replaces the original 38496 (TR105) has a different casing, of course... So we'll have to order adapted heat-sinks.

This is the usual problem when repairing Vintage amps or vintage cars the like: finding the replacement-parts ....

(Picture Copyright

Revision Step 4: Mission accomplished ;-)

Apparently the TO-5 Heatsinks don't exist anymore, my suppliers tell me... It's a problem we see more and more when looking for vintage components: "Product-lifecycles" get shorter and shorter.

With some heath-conducting glue I managed to put the new TR105 in with the old heatsink. I checked it with my Fluke thermometer for an hour playing disco @ maximum volume and it did fine (40° C).

I also had to change TR107 and MR105/106.

Then came the final calibration:

- The power-supply-voltage (with RV200) to exactly 67 Volt
- The output DC before the cap (with RV100) to exactly 33,5 Volt
- The rest-current to 33 mA (with RV101), this means 20 mV between pin 4 and 6 of the driver-board

After 2 hours of playing music @ maximum volume those calibrations were repeated.

The 303 sounds very transparent now and no trace of crossover-distortion can be measured on my scope.

Up to the next project...


PS. Joost added:

The transistor equivalents are as follows:
TR105 39496 BC461 2N5322
TR106 39495 BC441 2N5320

PPS. Chris (UK) says that the TO5 hetasinks are still available @


Anonymous Anonymous said...

Hello Stefaan,

Another 303 saved from the dustbin,very good. I got some extra information, maybe it will help you.

The transistor equivalents are as follows:
TR105 39496 BC461 2N5322
TR106 39495 BC441 2N5320
The quiescent current should be between 5 and 10 mA, this is collector current in Tr1! Measuring volage across pin 4 and 6 can give false results. I use this as a first indicator, then I measure the collector current.

Regards Joost Plugge

5:16 PM  
Blogger Stefaan said...

Hi, Joost,

According to Ruud Janssen (see
the best quiescent current is between 20 and 35 mA.

I measure the less cross-over distortion @ this value as well.

I'll post the equivalents above in the article.

Kind regards, Stefaan

7:03 PM  
Anonymous Anonymous said...

Hello Stefaan,

The value of 5 to 10 mA quiescent current in Tr1 is published in the official service documentation from Quad. If you use a value of 20 or 35 mA no harm can be done. I never heard or measured any differance, but if you measured less distortion, go for the higer value.

Regards Joost Plugge

5:44 PM  
Anonymous Anonymous said...

Distortion of this level is so less, it can hardly be measured by amateurequipment. In practice,if a scope image gives no *flat* trace at the sinus 0 position (between the pos and neg half) the is no cross over distortion.
In my opinion is 10 mA current more dan enough for de super beta triplets in the 303.


7:36 PM  
Anonymous EJP said...

But he says in that article that he has increased the 'standard 20mA' by 75% to 35mA. Where does he get the 'standard 20mA' from?

This is quite worrying.

(I don't speak Dutch so I am relying on a web translator.)

2:32 AM  
Blogger Stefaan said...

I guess he sees it on his scope.

Some 303's are OK with 20 mA, some others need 30...35 mA.

I guess it's the specs of the 2N3055's.


2:42 AM  
Blogger Stefaan said...

BTW, one of my Internet-friends proposed the BD183 as a better alternative for the 2N3055.

I'll check it out & keep you posted.


3:07 AM  
Anonymous EJP said...

OK, so you're saying that 20-35mA is for the replacement transistors, is that right? and 5-10mA is still right for the original ones - which would make the required voltage across R124/125 3-6mV.

Otherwise I would have to be very worried about a proposal to increase the bias current by a factor of 7.

7:52 AM  
Anonymous EJP said...

... and in fact too much bias current will convert this amplifier into a class AB which has *more* distortion, and also runs hotter, possibly too hot for the heatsinks. I would be very wary about all this.

8:35 AM  
Anonymous Anonymous said...

Hello, It is very difficult to determine the right quiescent current from the datasheet of the transistors. In the case of the 303 output circuit, the quality of the triplet is more important than the quality of the single power transistor. So a useful indicator could be the collector cut off current, this is the minimum collector current to keep the transistor in conducting mode. For a 2N3055 this is 5mA max, thus the quiescent current in the Quad service manual, 5 to 10mA, seems ok. The replacement transistor MJ15003 has a cut off current of 2mA max, so the value of 5 to 10 mA still seems ok. A higher value, 50mA, won't harm the 303. It is still very low comparing to the current when a signal is applied. If it is useful? Very difficult to measure, a 303 have distortion figures in the range of 0,01 %, so you need very sophisticated equipment and carefully setup the measurement to get meaningful results.

Regards Joost Plugge

1:53 PM  
Blogger Nunes said...

Hi Stefaan

I've replaced Tr105 & 106 to BD 140 & 139, with 100pf Miller cap on TR105 base & collector, with very good results. Cap C100 changed to polypropylene ans TR 100 changed to 2N 5087. Best results can be obtained on quiescent current adjustment by following the net-audio tips at

Best regards

Paulo Nunes

5:07 AM  
Anonymous Anonymous said...

Hello, Measuring the voltage across R124, R125 or both, based on the value of the wanted quiescent current can give false results, or not? Net Audio calculates 5mV across R125 or R125 giving 16.6 mA DC current trough R124. According to Quad 5-10mA current trough Tr1 is enough. But....through R124 runs also the DC current flowing from Tr109 and Tr105. You can measure the DC voltage across R120 and R123 and calculate the current. These currents are, 0.8mA and 8.6mA. This will give a bias current through Tr1 of 7.2 mA. So the total DC current should be between 14.4 - 19.4 mA, giving a DC voltage across R124 of 4.3 - 5.8 mV to meet the Quad specification. More accurate is to measure the voltage across R124 and R125 (leads 4 and 6 on the driver boards). It doubles the value of the voltage, but with some voltmeters this give better results. I still prefer measuring the real current by breaking lead two on the driver board. But, the Net Audio way will do the job.
Regards Joost Plugge

6:06 PM  
Anonymous Maximino Gonzalez said...

Hello Stefaan, I'm Maximino Gonzalez from Mexico City, fortunately have found you while surfing in the web looking for info about how to repair a 303 (serial number 13874) with similar problem as Peter's, my problem is I'm not in Europe so, can't send the 303 for be repaired by you.
Well, the case is I have this 303 since 1970, (I bought it second hand from someone who brougth it from Japan) and enjoyed it a lot up to the last week when one of the power supply capacitors had blown.
It is very dificult to get parts here in Mexico for repairing the amplifier, so I wander whether you can sell to me a new complete power supply card, send it to me and tell me the price for this.
I own a pair of quad ELS57, refurbished by myself, one 44-405 set and another 33-FM3-303 and proudly enjoy listening them playing all kind of music.

Best regards.

Maximino Gonzalez

7:26 AM  
Blogger Stefaan said...

Hallo, Maximino,

Please send me an e-mail @

We'll find a solution for the blown up power suplly in the 303 but I'll need your e-mail adress to discuss this.


9:17 AM  
Anonymous Anonymous said...

Hello Maximino,

Look on Ebay, I sell driver and Psu boards for the 303. If you wish I can also diliver a compleet Psu board. Contact me. But are you shure that's the only problem? Otherwise you are spending a lot of money and time with no results. Ebay name joost456.

Regards Joost Plugge

9:23 AM  
Anonymous Maximino Gonzalez said...

Hello Joost, as already said Stefaan, I ordered the PSU from NET Audio so thanks for the offering, I'll get you informed

Kind Regards


7:54 AM  
Anonymous Anonymous said...

I've just rebiased my Quad with the aid of a distortion analyser (AA501 with SG505 and scope). In my opinion Quad's recommendation of 5 to 10mA is entirely correct.

The 20-35mA recommendation seen here is much too high: it shows significant gm-doubling crossover distortion, and will also cause the output stage to run much hotter than it is designed for.

The midrange of Quad's recommendation, i.e. 7.5mA (4.5mV across pins 4 and 6) is just about right for minimum crossover distortion and THD.

7:22 AM  
Anonymous Anonymous said...

I seem to remember a way to adjust the quiescent current as publicized by Transtec a long time ago. Shorten the input of the channel to be adjusted to earth. Feed a 1000 Hz signal to the other channel. Connect the output from this channel via a 120 Ohm resistor into the output of first channel. Connect a scope across the input of the first channel. Adjust until the crossover effect is nil or minimum. It worked for me at the time. The trick depends on the fact that the damping factor rises sharply at a maladjusted crossover point. Caveat: I don't know for sure whether 120 Ohm is the right value.

9:07 PM  
Blogger EJP said...

Something wrong with the description for that trick. If the input is shorted how can it shown anything on a scope?


3:45 AM  

Post a Comment

Links to this post:

Create a Link

<< Home