Hi,
Sorry i posted this at the end of a very long thread so it wasn't really seen, so thought i'd start a thread
I have 2 cambridge audio A5i amps, these amps uses Sanken sap15 darlington pair transistors, 1 of which is dead (every leg seems to be shorted).
My question is, if i need to replace 1 sap15, is it better to get a complete set with the same batch number? i've noticed that there is some difference between the readings between the different saps in the two different players (things like checking diode readings which are currently identical to the other saps in that player, but slightly different to the other amp's.
I assume it will also mean i will probably have to adjust the variable resistors to get the idling current to as close to 40mA as possible, anything else that needs to be done if the new transistors are slightly different to the other channel?
Thanks,
Andy
Sorry i posted this at the end of a very long thread so it wasn't really seen, so thought i'd start a thread
I have 2 cambridge audio A5i amps, these amps uses Sanken sap15 darlington pair transistors, 1 of which is dead (every leg seems to be shorted).
My question is, if i need to replace 1 sap15, is it better to get a complete set with the same batch number? i've noticed that there is some difference between the readings between the different saps in the two different players (things like checking diode readings which are currently identical to the other saps in that player, but slightly different to the other amp's.
I assume it will also mean i will probably have to adjust the variable resistors to get the idling current to as close to 40mA as possible, anything else that needs to be done if the new transistors are slightly different to the other channel?
Thanks,
Andy
Hi Andy,
Oh my! Do you know anyone that can help you and show you a few things?
It's the small things that will get you! Try to feel how tight the mounting screws are. Clean the heatsinks, the mounting surfaces must be spotless. Don't use insulators in poor condition, use fresh grease or "plastic compound" insultors.
-Chris
Oh my! Do you know anyone that can help you and show you a few things?
It's the small things that will get you! Try to feel how tight the mounting screws are. Clean the heatsinks, the mounting surfaces must be spotless. Don't use insulators in poor condition, use fresh grease or "plastic compound" insultors.
-Chris
Hi Andrew
I have made amplifiers before, but they have all been pretty simple chip amps, so I've done alot of readig on power supplies, decoupling, snubbers etc.. I should also be fine with the "small things", i have a new tube of decent thermal grease and even some good mylar insulators.
It's about time i look into the world of solid state amplifiers anyway
Thanks again for the help
Andy
I have made amplifiers before, but they have all been pretty simple chip amps, so I've done alot of readig on power supplies, decoupling, snubbers etc.. I should also be fine with the "small things", i have a new tube of decent thermal grease and even some good mylar insulators.
It's about time i look into the world of solid state amplifiers anyway
Thanks again for the help
Andy
elaar said:...it looks like i'm going to have to read up a little as I have no idea what the output offset is, let alone measure it and adjust
That's the easy one, it's the DC voltage across the speaker terminals at zero input and should be zero.
The methods are given here though for idle current I'd stick to the maker's recommendation; commercial amps are unlikely to have the heatsinking to cope with the high levels he advocates.
Hi, thanks for the reply cpemma.
That's a very helpful tutorial. Without access to the appropriate cambridge audio service manual I wouldn't like to hazard a guess to the original quiescent currents, each channel uses a pair of SAP15s, would anyone like to guess a likely value?
It's a shame the channels aren't completely seperated with different bridges because at least then 1 channel would still be working and i could take readings from that.
Thanks,
Andy
That's a very helpful tutorial. Without access to the appropriate cambridge audio service manual I wouldn't like to hazard a guess to the original quiescent currents, each channel uses a pair of SAP15s, would anyone like to guess a likely value?
It's a shame the channels aren't completely seperated with different bridges because at least then 1 channel would still be working and i could take readings from that.
Thanks,
Andy
Matching is important for VOS
Yes,
AndrewT is very correct. Matching them for hFE (which for Darlingtons can be in the 1000 - 5000 or even higher range) is very important on many levels.
Class B elimination of 2nd order harmonic distortion can't occur unless the matching is very close.
And the VOS (voltage offset when the inputs are shorted) is also a big concern. Even a relatively small difference in hFE can cause the output to very strongly favor either the positive or negative side of the waveform. That's seriously bad for tone quality, and can even damage or burn out your woofers (the tweeters are protected from VOS / DC offset by the crossover capacitors).
Hope this helps.
Yes,
AndrewT is very correct. Matching them for hFE (which for Darlingtons can be in the 1000 - 5000 or even higher range) is very important on many levels.
Class B elimination of 2nd order harmonic distortion can't occur unless the matching is very close.
And the VOS (voltage offset when the inputs are shorted) is also a big concern. Even a relatively small difference in hFE can cause the output to very strongly favor either the positive or negative side of the waveform. That's seriously bad for tone quality, and can even damage or burn out your woofers (the tweeters are protected from VOS / DC offset by the crossover capacitors).
Hope this helps.
Hi,
I was thinking of matching the Vbe of the darlingtons not the hFE.
But, I see now that there are only ONE pair per channel and Vbe cross match between NPN and PNP is not quite so important, but still worth doing.
I do not know of the argument that mismatch between hFE across the N & P pair causes an excess of distortion, maybe I need to do more reading.
I was thinking of matching the Vbe of the darlingtons not the hFE.
But, I see now that there are only ONE pair per channel and Vbe cross match between NPN and PNP is not quite so important, but still worth doing.
I do not know of the argument that mismatch between hFE across the N & P pair causes an excess of distortion, maybe I need to do more reading.
Hi, thanks for all of the helpful replies, much appreciated.
There is only 1 trimmer pot per channel, this must adjust the quiescent current I'm assuming, so how exactly is the DC offset adjusted? or does this change as an indirect result of adjusting the current?
Also, i have replaced the other damaged parts that were affected during the fault (resistors, a few diodes etc..). The trim pots looked okay but it wasn't until i took some readings from them I found they must have also been damaged, 100ohm pots (100222M), yet one reads 180ohms and the other 50ohms (measured from one end to the other (not the slider)). Is this how faulty/overpowered trim pots tend to measure?
Many thanks,
Andy
There is only 1 trimmer pot per channel, this must adjust the quiescent current I'm assuming, so how exactly is the DC offset adjusted? or does this change as an indirect result of adjusting the current?
Also, i have replaced the other damaged parts that were affected during the fault (resistors, a few diodes etc..). The trim pots looked okay but it wasn't until i took some readings from them I found they must have also been damaged, 100ohm pots (100222M), yet one reads 180ohms and the other 50ohms (measured from one end to the other (not the slider)). Is this how faulty/overpowered trim pots tend to measure?
Many thanks,
Andy
Not Theoretical, Been There
My comments about the hFE matching are not theoretical in nature. They are based on actual output device replacements in consumer electronics amplifiers over a period of over 40 years. I screwed up my first amplifier output device replacement in 1969, on an Allied Radio product that a friend of mine had built from a kit. He came to me when he smoked it, and asked me to help him fix it. It took two more rounds of smoke before I figured out that the hFE of the replacement part had to match the hFE of the existing complementary pair.
It was a hard lesson, but I've never forgotten it. Later, another friend brought me a Dynaco Stereo 80 with an output device that was destroyed. He had purchased a spot on replacement part and bolted it to the heat sink, only to find that the channel ran extremely hot and was distorted. The reason? The hFE of the replacement was not a match for the "good" transistor that was still remaining.
Here's the problem with depending on the feedback loop to correct for this. When the hFE doesn't match, the "weaker" transistor has to be driven all the time. It might as well be running in Class A, because the DC Servo or DC feedback has to "push" that side of the rail back to where it is supposed to be.
That means that all of the drive circuits on that side of the waveform are being driven all the time in order to eliminate the VOS. That's fine when the amplifier is at rest, but when a real music signal comes through, now half the waveform is pushing one entire side of the amplifier harder than the other. The result is that the driver circuit goes into overdrive on the "weak" side at a much lower volume than it would normally. In the case of a fairly obvious mismatch in hFE, the driver circuit can be overdriven at normal volume levels.
You can try it if you like, but I've already been there and done that. I wish you a lot of luck with it. Uh, put some nice fuses in your speaker cables, say 3/4 amp normal speed (never slow blow) or less. You know, just in case I'm right.
My comments about the hFE matching are not theoretical in nature. They are based on actual output device replacements in consumer electronics amplifiers over a period of over 40 years. I screwed up my first amplifier output device replacement in 1969, on an Allied Radio product that a friend of mine had built from a kit. He came to me when he smoked it, and asked me to help him fix it. It took two more rounds of smoke before I figured out that the hFE of the replacement part had to match the hFE of the existing complementary pair.
It was a hard lesson, but I've never forgotten it. Later, another friend brought me a Dynaco Stereo 80 with an output device that was destroyed. He had purchased a spot on replacement part and bolted it to the heat sink, only to find that the channel ran extremely hot and was distorted. The reason? The hFE of the replacement was not a match for the "good" transistor that was still remaining.
Here's the problem with depending on the feedback loop to correct for this. When the hFE doesn't match, the "weaker" transistor has to be driven all the time. It might as well be running in Class A, because the DC Servo or DC feedback has to "push" that side of the rail back to where it is supposed to be.
That means that all of the drive circuits on that side of the waveform are being driven all the time in order to eliminate the VOS. That's fine when the amplifier is at rest, but when a real music signal comes through, now half the waveform is pushing one entire side of the amplifier harder than the other. The result is that the driver circuit goes into overdrive on the "weak" side at a much lower volume than it would normally. In the case of a fairly obvious mismatch in hFE, the driver circuit can be overdriven at normal volume levels.
You can try it if you like, but I've already been there and done that. I wish you a lot of luck with it. Uh, put some nice fuses in your speaker cables, say 3/4 amp normal speed (never slow blow) or less. You know, just in case I'm right.
I forgot about the Vbe
Uh, I'm sorry, I forgot about your comment on the Vbe. That's a common error to make.
Vbe does not have to be matched in output devices, because it drops like a stone as soon as the devices heat up. At room temperature, you will probably have double diode values of around 1.5 volts from the Darlingtons. But that is so temperature dependent that you can test it with your fingers.
Set the test leads from your forward voltage drop meter on one of the devices. Now, pick up the device and press your fingers onto both of the flat sides. Wait a moment for your body temperature to heat the device. You should see your Vbe start to drop. The longer you hold the device, the farther the Vbe will drop!
Some designers have a "sensing diode" that is mounted to the heat sink in order to detect the temperature of the output devices. As the forward voltage drop of the diode goes down it sends a signal to the circuit to reduce the static bias current. This prevents the bias from running away as the devices heat up.
Uh, I'm sorry, I forgot about your comment on the Vbe. That's a common error to make.
Vbe does not have to be matched in output devices, because it drops like a stone as soon as the devices heat up. At room temperature, you will probably have double diode values of around 1.5 volts from the Darlingtons. But that is so temperature dependent that you can test it with your fingers.
Set the test leads from your forward voltage drop meter on one of the devices. Now, pick up the device and press your fingers onto both of the flat sides. Wait a moment for your body temperature to heat the device. You should see your Vbe start to drop. The longer you hold the device, the farther the Vbe will drop!
Some designers have a "sensing diode" that is mounted to the heat sink in order to detect the temperature of the output devices. As the forward voltage drop of the diode goes down it sends a signal to the circuit to reduce the static bias current. This prevents the bias from running away as the devices heat up.
Hi Chewrock,
I have read your post16, only once so far, and do not believe you have analysed the problem correctly. Unfortunately I do not have the experience to give you a reasoned counter argument.
Post 17 is close to right but
I have read your post16, only once so far, and do not believe you have analysed the problem correctly. Unfortunately I do not have the experience to give you a reasoned counter argument.
Post 17 is close to right but
is completely untrue. Just look at the manufacturers' data and curves to see the variation with temperature.
Couple of thoughts you may rip to pieces at your leisure:
1. Darlington's have such high gain that a real 'match' is gonna be nigh-on impossible.
2. I don't have 40 years experience doing the amp repair thing, but in a standard complementary output stage using a common collector topology, swamping the emitter with a low value resistance is usually plenty good enough to compensate for gain variations in the output devices. Matched devices are nice, but not very necessary with this topology. mikeks has already more or less said the same thing.
3. Variations in Vbe are significant enough to use for bias compensation, but in output devices who cares what Vbe is doing?
1. Darlington's have such high gain that a real 'match' is gonna be nigh-on impossible.
2. I don't have 40 years experience doing the amp repair thing, but in a standard complementary output stage using a common collector topology, swamping the emitter with a low value resistance is usually plenty good enough to compensate for gain variations in the output devices. Matched devices are nice, but not very necessary with this topology. mikeks has already more or less said the same thing.
3. Variations in Vbe are significant enough to use for bias compensation, but in output devices who cares what Vbe is doing?
elaar said:
My NAD3020 only has one trimpot per channel, for DC-Offset adjust.
Quiescent current is checked across the emitter resistor as usual and, if not in range, a fixed resistor in parallel with another (lower value) resistor is changed to suit.
Shown here as R641 (Rx) in parallel with R647 on Q609 base.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.