I repaired one of these many years ago. It kept blowing output transistors. With a current limited power supply I was eventually able to trace it to an open bias trim pot which had a crack in the carbon. So I recommend that as part of the upgrade that you should replace the bias trim pots 
regards
Trevor
regards
Trevor
The BD139/140 are marginal with regard to VCE rating where the phenomena of secondary breakdown comes into play.
In practice the MJE devices will be fine, particularly given the hfe of the more modern 2N3773
Good advice... ancient pots can be troublesome.
In practice the MJE devices will be fine, particularly given the hfe of the more modern 2N3773
I repaired one of these many years ago. It kept blowing output transistors. With a current limited power supply I was eventually able to trace it to an open bias trim pot which had a crack in the carbon. So I recommend that as part of the upgrade that you should replace the bias trim pots
regards
Trevor
Good advice... ancient pots can be troublesome.
Can't agree with you there is simply not enough current available to handle a short or even low impedance loads with the .5A devices. The BD's are off by less than 10% should be fine given typical margin for newer devices if the OP is concerned he could use 2SD669/2SB649 which would be an excellent choice or pretest/select the BD for breakdown voltage.
Have to agree to disagree on that one
You can't assume a device will have better specs than the data sheet says. Without any VI limiting and SOA protection, failure with a short seems inevitable, and the scenario that caused this very problem.
Just out of interest I looked up Doug Selfs "load invariant" design and that too used MJE340/350's despite driving parallel output pairs. I'd be quite happy with the MJE's
The 2SD/2SB's you mention are a good choice too.
Not familiar with those but would almost certainly be OK...
I have suggested and used these in the past but on closer inspection they have worse SOA than the BDs even with their 4A max spec, they would probably work fine and do at least meet the voltage spec.
Have to agree to disagree on that one
Right say for a Military or new design, but statistically for this particular spec it should be fine.
The problem was that it had the wrong fuses in it, currents will soar with a short and the fuse can take about 1 second to blow so you need good current margin and particularly SOA in the drivers.
They are not a good part for this application and drivers in general.
Last edited:
At the minute I am almost settled on STMicroelectronics 2N3773's, Central 2N5322 and 2N5320's from Mouser, and STMicroelectronics BF259's and the resistors I need from Farnell.
I'm waiting on the arrival of a new soldering iron so I can test the rest of the components surrounding before I place the orders (postage is a bitch from both companies!).
What would be the reason behind replacing with different drivers? Surely if I can find like components to the originals that would be the best bet?
Scream now if you don't like what you see, I'm still learning.
I'm waiting on the arrival of a new soldering iron so I can test the rest of the components surrounding before I place the orders (postage is a bitch from both companies!).
What would be the reason behind replacing with different drivers? Surely if I can find like components to the originals that would be the best bet?
Scream now if you don't like what you see, I'm still learning.
Last edited:
Perfect choice.
Reason behind me suggesting different drivers was to enable you to order all from one company, and tbh I didn't scour different companies to see if the originals were available
The design of the amp is a classic "text book" circuit and substituting parts would present no problem.
Make sure the mica washers on the outputs are OK. After all these years they may fall to pieces.
When you have replaced the parts remember you MUST use a bulb tester as that will save any major failures should anything still be faulty.
Reason behind me suggesting different drivers was to enable you to order all from one company, and tbh I didn't scour different companies to see if the originals were available
The design of the amp is a classic "text book" circuit and substituting parts would present no problem.
Make sure the mica washers on the outputs are OK. After all these years they may fall to pieces.
When you have replaced the parts remember you MUST use a bulb tester as that will save any major failures should anything still be faulty.
Another thing that has made me think...
I have read somewhere that when it comes to the output caps you can basically put as much capacitance on as you like (within reason). The original caps in the amp were 4000uf and I bought 4700uf's as replacements. I have also read that caps can be wired in parallel to add together their capacitance, and as the caps I bought came in a pack of 4 I opted to wire a pair in on each channel, bringing my total up to 9400uf on each channel.
This wont be putting added strain on the amplifier ... will it? (a niggling doubt in my head is saying "think of the amp having to drive the caps as well...")
I have read somewhere that when it comes to the output caps you can basically put as much capacitance on as you like (within reason). The original caps in the amp were 4000uf and I bought 4700uf's as replacements. I have also read that caps can be wired in parallel to add together their capacitance, and as the caps I bought came in a pack of 4 I opted to wire a pair in on each channel, bringing my total up to 9400uf on each channel.
This wont be putting added strain on the amplifier ... will it? (a niggling doubt in my head is saying "think of the amp having to drive the caps as well...")
At the minute I am almost settled on STMicroelectronics 2N3773's, Central 2N5322 and 2N5320's from Mouser, and STMicroelectronics BF259's and the resistors I need from Farnell.
I'm waiting on the arrival of a new soldering iron so I can test the rest of the components surrounding before I place the orders (postage is a bitch from both companies!).
What would be the reason behind replacing with different drivers? Surely if I can find like components to the originals that would be the best bet?
Scream now if you don't like what you see, I'm still learning.
I'd prefer Onsemi for the 2N3773s but those should do, yes originals on the drivers are best as long as they are of good quality. I'd test all the semis in the power amp section just to be sure and not have to order more parts. Obviously fakes can be an issue also.
You can also use a current limiting resistor across the fuse holder to protect the new devices while you test the amp.
Last edited:
Farnell and Mouser come across as fairly straight-up companies, are they likely to serve fakes?! I suppose we should have learnt by now if you can't trust pharmaceuticals...
Central don't come across as the very best in their class, I get the feeling that they are to semi's what JVC are to hi(lo)-fi. But still, the only company that appear to be selling 2N5322/20's in this country. :/
Central don't come across as the very best in their class, I get the feeling that they are to semi's what JVC are to hi(lo)-fi. But still, the only company that appear to be selling 2N5322/20's in this country. :/
Sonically I suspect you won't gain much. The cap presents an increasing impedance as frequency falls... in other words the bass is rolled off.
The reactance of the cap is given by 1/2piFC and 4700 uf is certainly large enough. The roll of also depends on the "reactance" of the load and a speaker is anything but a constant 4 or 8 ohm load. And in an AC coupled amp like this it's one of many such compromises.
Will the amp suffer... no
However those caps have to charge, and they do that via the speakers so it's those that will complain most. Depending how well the amp is behaved at switch on and switch off you may notice a larger "thump" and more movement of the bass driver cone at these times.
I always let the amp charge up fully for a good 5 minutes before I engage the speakers just to be on the safe side, so that isn't a concern. Thanks for the tip anyway...
The amp is not particularly bass-heavy, so I might contemplate taking the second pair out depending on how it sounds when it's up and running. Does that sound like a good idea to you?
As the amp stands 4700uf is plenty IMO. Unless you are running it through speakers with a very extended LF response and in a room that can support that then there is just no point increasing the value. Even then I think you would have to look carefully at making sure no other cap values were a limiting factor.
It's dead easy to rig a relay up to delay the speaker connection and give a fairly quick turn off.
The 3300 resistor after the cap gives a charging time of T=CR which is around 16 seconds to bring the cap to around 63%
We call that time period 1T. However it takes progressively longer for the cap to fully charge (it follows an exponential curve). 4T (about a minute) brings it too around 99%.
You could usefully decrease the value of the 3300 ohm to around an 820 ohm 1watt.
That would reduce T1 to under 4 seconds.
It's dead easy to rig a relay up to delay the speaker connection and give a fairly quick turn off.
The 3300 resistor after the cap gives a charging time of T=CR which is around 16 seconds to bring the cap to around 63%
We call that time period 1T. However it takes progressively longer for the cap to fully charge (it follows an exponential curve). 4T (about a minute) brings it too around 99%.
You could usefully decrease the value of the 3300 ohm to around an 820 ohm 1watt.
That would reduce T1 to under 4 seconds.
Ok, new soldering iron in hand...
I have now gone on to test the smaller semi's in the circuit that I wasn't able to before - BC267/BC297 - and they both came out as faulty according to...
Testing semiconductors with analog and digital multimeters
I decided to test the semi's on the starboard (and fully functional) channel, to make sure I was finding what I was looking for. I wanted them to read out at what this website describes as a working transistor, as they are working...
They didn't... they read out as exactly the same as their counterparts in the damaged channel.
HOW CAN THIS BE?! Is the website wrong or is it just me being a complete idiot? I am using the diode check setting on a "Sinometer M-830B" digital multimeter.
I have now gone on to test the smaller semi's in the circuit that I wasn't able to before - BC267/BC297 - and they both came out as faulty according to...
Testing semiconductors with analog and digital multimeters
I decided to test the semi's on the starboard (and fully functional) channel, to make sure I was finding what I was looking for. I wanted them to read out at what this website describes as a working transistor, as they are working...
They didn't... they read out as exactly the same as their counterparts in the damaged channel.
HOW CAN THIS BE?! Is the website wrong or is it just me being a complete idiot? I am using the diode check setting on a "Sinometer M-830B" digital multimeter.
Probably down to your meter I guess. What readings do you get ?
It's assumed that your meter on ohms and diode ranges actually has a "positive" voltage on the red lead but it's not guaranteed alhough I have never come across a DVM that wasn't like that. Older analogue meters commonly had the "positive" test voltage on the black lead.
The diode check on the DVM normally shows the forward volt drop of the juntion in millivolts so I would expect to see something like 0.657 (for example) one way and whatever is normal for the meter with nothing connected when the leads are reversed.
Post your results for B to E measured one way and same again measured with the leads reversed.
Also you must do these measurements with the device isolated (out of circuit) as other components influence the readings. The slightest residual voltage (from the caps in the PSU) will give totally wild readings.
It's assumed that your meter on ohms and diode ranges actually has a "positive" voltage on the red lead but it's not guaranteed alhough I have never come across a DVM that wasn't like that. Older analogue meters commonly had the "positive" test voltage on the black lead.
The diode check on the DVM normally shows the forward volt drop of the juntion in millivolts so I would expect to see something like 0.657 (for example) one way and whatever is normal for the meter with nothing connected when the leads are reversed.
Post your results for B to E measured one way and same again measured with the leads reversed.
Also you must do these measurements with the device isolated (out of circuit) as other components influence the readings. The slightest residual voltage (from the caps in the PSU) will give totally wild readings.
This is all the readings I made of all the Semi's out of circuit. There are some interesting figures here. I really don't know what to make of it! I can't see how the right channel can be working fine if the components are reading out completely bizarrely...
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.