Fin,
The later version of the amp (Alpha9) uses an LM393 comparator which compares the voltage across the 0R22 resistors in the output stage with a reference level and feedsback to adjust the current source in the long-tail pair; thus adjusting the bias. Anyway to see if you have this or not look for the LM393 IC. It should somewhere between the large heatsink and the toroid. If you don't have this then there should be a pot somewhere on the pcb to adjust the bias for each channel. The instructions are to initially set the voltage across the 0R22 resistors (R1,R101) to 5mV. Load the output with an 8R load and apply 1W rms 1KHz sine wave. Wait for 15 minutes (for the temperature to stablise) and then re-set the voltage across the 0R22 resistors to 4mV.
As far as changing other things goes I did start out with the idea of tweaking this amp but as I got further and further into simulating my modifications I eventually ended up with a radically different design (which I have yet to build btw). You can do the same things that are recommended for any tweaking project though.
Look at the power supply, change the diodes for soft fast recovery types, experiment with snubber caps across the diodes and after the toroid.
Add a filter to the mains input. Something like a few nF between live-earth and neutral-earth (mains rated!) should do. Try and avoid inductors so don't use a filtered IEC socket although you could try passing the mains cables through a ferrite core (1 loop). You could also put a shield around the psu / toroid and screen the power cables. The routing of the supply from the toroid to the bridge diodes at the centre of the pcb isn't ideal!
Look at the ground return paths from each stage of the amplifier to the supply (you don't want the large currents flowing in the output stage to cause ground bounce on the pre-amp supplies!).
Look at decoupling and reducing noise on the +/-15V supply and at the regulation circuit (currently standard LM317/LM337).
Change the DC servo op-amp (IC1,101) for one with a faster slew rate and lower offset.
Change/remove the DC coupling cap (C12,C112).
Replace the 10000uF reservoir caps for something larger (perhaps 22000uF) and something smaller (1000-4700uF) and bypass these with a smaller (100nF) film cap (and perhaps a ceramic).
Improve local decoupling on the opamps.
Think about changing the opamps in the pre-amp for high spec ones (currently TL072 and NE5534 if memory serves me correctly).
I'm sure there are more things you could fiddle with but I'm sure that's enough to be going on with !
Cheers,
Jon
The later version of the amp (Alpha9) uses an LM393 comparator which compares the voltage across the 0R22 resistors in the output stage with a reference level and feedsback to adjust the current source in the long-tail pair; thus adjusting the bias. Anyway to see if you have this or not look for the LM393 IC. It should somewhere between the large heatsink and the toroid. If you don't have this then there should be a pot somewhere on the pcb to adjust the bias for each channel. The instructions are to initially set the voltage across the 0R22 resistors (R1,R101) to 5mV. Load the output with an 8R load and apply 1W rms 1KHz sine wave. Wait for 15 minutes (for the temperature to stablise) and then re-set the voltage across the 0R22 resistors to 4mV.
As far as changing other things goes I did start out with the idea of tweaking this amp but as I got further and further into simulating my modifications I eventually ended up with a radically different design (which I have yet to build btw). You can do the same things that are recommended for any tweaking project though.
Look at the power supply, change the diodes for soft fast recovery types, experiment with snubber caps across the diodes and after the toroid.
Add a filter to the mains input. Something like a few nF between live-earth and neutral-earth (mains rated!) should do. Try and avoid inductors so don't use a filtered IEC socket although you could try passing the mains cables through a ferrite core (1 loop). You could also put a shield around the psu / toroid and screen the power cables. The routing of the supply from the toroid to the bridge diodes at the centre of the pcb isn't ideal!
Look at the ground return paths from each stage of the amplifier to the supply (you don't want the large currents flowing in the output stage to cause ground bounce on the pre-amp supplies!).
Look at decoupling and reducing noise on the +/-15V supply and at the regulation circuit (currently standard LM317/LM337).
Change the DC servo op-amp (IC1,101) for one with a faster slew rate and lower offset.
Change/remove the DC coupling cap (C12,C112).
Replace the 10000uF reservoir caps for something larger (perhaps 22000uF) and something smaller (1000-4700uF) and bypass these with a smaller (100nF) film cap (and perhaps a ceramic).
Improve local decoupling on the opamps.
Think about changing the opamps in the pre-amp for high spec ones (currently TL072 and NE5534 if memory serves me correctly).
I'm sure there are more things you could fiddle with but I'm sure that's enough to be going on with !
Cheers,
Jon
Hi Jon
Thanks for all this detail
I will be looking for the LM393 this weekend. Hopefully it is there as I don't think I have the ability to apply 1W rms 1KHz sine wave. Does this require a signal generator?
Are you referring to D201/2/3/4? These are MUR420 Ultrafast diodes. What would be a good replacement?
I'll have to ask you about this when I take the pcb out.
Would you retain the LM317/337?
Any recomendations? OPA627?
Should each 10,000uF cap be replaced with a 22,000uF and a (1000-4700uF) cap?
Again - would OPA627 be good here?
Should I be thinking of replacing the SC2547 transistors - in case they have been damaged?
Is it worthwhile disconnecting/disabling some of the features that I don't require, such as:- the speaker swiches, hardwire a bypass of the tone controls....?
Thanks again
Fin
Thanks for all this detail
I will be looking for the LM393 this weekend. Hopefully it is there as I don't think I have the ability to apply 1W rms 1KHz sine wave. Does this require a signal generator?
JonHarrison said:
Are you referring to D201/2/3/4? These are MUR420 Ultrafast diodes. What would be a good replacement?
JonHarrison said:
I'll have to ask you about this when I take the pcb out.
JonHarrison said:
Would you retain the LM317/337?
JonHarrison said:
Any recomendations? OPA627?
JonHarrison said:
Should each 10,000uF cap be replaced with a 22,000uF and a (1000-4700uF) cap?
JonHarrison said:
Again - would OPA627 be good here?
JonHarrison said:
Should I be thinking of replacing the SC2547 transistors - in case they have been damaged?
Is it worthwhile disconnecting/disabling some of the features that I don't require, such as:- the speaker swiches, hardwire a bypass of the tone controls....?
Thanks again
Fin
Hi Fin,
Yes you will require a signal generator to set the bias properly.
The MUR420 are ultrafast recovery diodes but I'm not sure if they have a soft recovery characteristic or not. They're probably fine but I've made a note to investigate UG10DCT , BYQ28E , BYW98-200 or SBYV28-200 alternatives. Look with a scope for any oscillation during their recovery period though (the diode capacitance and transformer inductance creates a resonance) and think about snubbing them. If it's possible to split the secondaries so that you have seperate 0-44 , 0-44 then you could fit a second bridge - at the moment there are only three wires coming from the toroid and not four. I'm not sure if you'd find that the common wire is simply linked if you unwrap the insulation a bit or whether there is a single winding and the centre (common) wire is simply tapped off half way.
I was thinking of replacing the LM317/337 either with different regulators, lower noise versions of the LM317/337 or a discrete regulator (perhaps keeping them in place for use as a pre-regulator). A quieter supply will help lower the noise floor and help reveal more of the subtle details in the sound.
As far as changing op-amps, look at the usual suspects OPA2134/2132/2604/627/634 , AD8610/8620/8065/8066/825 , LM6172 , LT1115.
For the servo I was going to look at AD8620 , OP177 or AD711.
I read somewhere that someone had replaced the TL072 and NE5534 with OPA604 and changed the servo op-amp to OPA134.
The servo introduces an unequal amount of feedback over the frequency range which is why some people recommend using a high quality (polypropylene) capacitor to block the DC instead. With the way this amp is designed though you need both, the coupling cap blocks any DC from the pre-amp and the dc-servo cancels any offsets in the power amp. You could possibly remove both and couple the output through a capacitor but I'm not sure I'd recommend that !
Try replacing the 2 x 10000uF with 4 x 10000uF or 2 x 22000uF (or even 8 x 4700uF) 63V with a low ESR and high ripple current. If you go above 10000uF also fit the smaller caps (1000-4700uF) so that the transient response doesn't suffer and what ever you do bypass the electrolytics with 100nF film for HF decoupling. You could also consider taking the psu out of the case to a seperate box and running the regulated +/-44V supplies to the amp with the bulk of the capacitance on the psu but also 1000-4700uF on the amp pcb.
You could replace the pre-drivers (2SC2547 and 2SA1085) but I'd try and check them first.
>Is it worthwhile disconnecting/disabling some of the features that I don't require, such as:- the speaker swiches, hardwire a bypass of the tone controls....?
You could try removing the PIC microcontroller, that might put a bit of noise on the supply (you'll loose the remote control); although I think it's fed from a seperate winding so you probably won't hear the diffence. The tone direct switch bypasses the tone circuitry with the signal only passing through a single TL072 buffer before the power-amp stage. There is no doubt that bypassing the tone control like this improves the sound. I don't know if anything could be done to the output relays. Also I don't know what if any crosstalk there is on the input selector switch (it may be worth grounding the unused input channels).
Cheers,
Jon
Yes you will require a signal generator to set the bias properly.
The MUR420 are ultrafast recovery diodes but I'm not sure if they have a soft recovery characteristic or not. They're probably fine but I've made a note to investigate UG10DCT , BYQ28E , BYW98-200 or SBYV28-200 alternatives. Look with a scope for any oscillation during their recovery period though (the diode capacitance and transformer inductance creates a resonance) and think about snubbing them. If it's possible to split the secondaries so that you have seperate 0-44 , 0-44 then you could fit a second bridge - at the moment there are only three wires coming from the toroid and not four. I'm not sure if you'd find that the common wire is simply linked if you unwrap the insulation a bit or whether there is a single winding and the centre (common) wire is simply tapped off half way.
I was thinking of replacing the LM317/337 either with different regulators, lower noise versions of the LM317/337 or a discrete regulator (perhaps keeping them in place for use as a pre-regulator). A quieter supply will help lower the noise floor and help reveal more of the subtle details in the sound.
As far as changing op-amps, look at the usual suspects OPA2134/2132/2604/627/634 , AD8610/8620/8065/8066/825 , LM6172 , LT1115.
For the servo I was going to look at AD8620 , OP177 or AD711.
I read somewhere that someone had replaced the TL072 and NE5534 with OPA604 and changed the servo op-amp to OPA134.
The servo introduces an unequal amount of feedback over the frequency range which is why some people recommend using a high quality (polypropylene) capacitor to block the DC instead. With the way this amp is designed though you need both, the coupling cap blocks any DC from the pre-amp and the dc-servo cancels any offsets in the power amp. You could possibly remove both and couple the output through a capacitor but I'm not sure I'd recommend that !
Try replacing the 2 x 10000uF with 4 x 10000uF or 2 x 22000uF (or even 8 x 4700uF) 63V with a low ESR and high ripple current. If you go above 10000uF also fit the smaller caps (1000-4700uF) so that the transient response doesn't suffer and what ever you do bypass the electrolytics with 100nF film for HF decoupling. You could also consider taking the psu out of the case to a seperate box and running the regulated +/-44V supplies to the amp with the bulk of the capacitance on the psu but also 1000-4700uF on the amp pcb.
You could replace the pre-drivers (2SC2547 and 2SA1085) but I'd try and check them first.
>Is it worthwhile disconnecting/disabling some of the features that I don't require, such as:- the speaker swiches, hardwire a bypass of the tone controls....?
You could try removing the PIC microcontroller, that might put a bit of noise on the supply (you'll loose the remote control); although I think it's fed from a seperate winding so you probably won't hear the diffence. The tone direct switch bypasses the tone circuitry with the signal only passing through a single TL072 buffer before the power-amp stage. There is no doubt that bypassing the tone control like this improves the sound. I don't know if anything could be done to the output relays. Also I don't know what if any crosstalk there is on the input selector switch (it may be worth grounding the unused input channels).
Cheers,
Jon
Hi Jon
Looks like I'm going to be busy with this amp for a while
I'll look more closely at the diodes and try to get some soft recovery types. With only limited access to a scope, I might try to narrow down the selection before testing.
I've already got some other versions of LM317/337 and most of the opamps you recommend - which is good news.
From what you say - leaving the servo intact and using a polypropylene cap to block the DC - seems like the way to go.
The deciding factor for replacing the 2 x 10000uF caps will be space. I'll have a look later today and see what might fit.
How do you check the pre-drivers? Is there a test or do you just listen?
Thanks again.
Fin
Looks like I'm going to be busy with this amp for a while
I'll look more closely at the diodes and try to get some soft recovery types. With only limited access to a scope, I might try to narrow down the selection before testing.
I've already got some other versions of LM317/337 and most of the opamps you recommend - which is good news.
From what you say - leaving the servo intact and using a polypropylene cap to block the DC - seems like the way to go.
The deciding factor for replacing the 2 x 10000uF caps will be space. I'll have a look later today and see what might fit.
How do you check the pre-drivers? Is there a test or do you just listen?
Thanks again.
Fin
Fin,
>I'll look more closely at the diodes and try to get some soft recovery types.
The existing MUR420 diodes are probably fine. Many people swear by the MUR860 which is just a higher voltage / current version. You could just add a small cap across each, perhaps in series with a resistor. To calculate the snubber correctly you'd need to know the inductance of the secondary winding of the transformer (which I don't) but you might start experimenting with something of the order of 2 to 20nF.
>From what you say - leaving the servo intact and using a polypropylene cap to block the DC - seems like the way to go.
That's probably what I'll do. Change C12 and C102 for something like 2.2uF or more (I think they are Rubycon in a grey sleeve at the moment). You could also change C8 / C108 which form part of the DC-servo but since it isn't directly in the audio path I probably wouldn't bother, besides you'd never get a polypropylene large enough (to avoid rolling off LF) - it's currently a 100uF and that's not practical.
>The deciding factor for replacing the 2 x 10000uF caps will be space. I'll have a look later today and see what might fit.
I don't think that they laid out the pcb for the wider lead pitch of larger caps , the 22000uF will probably have screw terminals anyway so you'll have to be creative in mounting them.
>How do you check the pre-drivers? Is there a test or do you just listen?
The circuit won't operate correctly with the output MOSFETs removed but you could look at the waveforms on a 'scope or just test the pre-drivers (standard NPN and PNP transistors) with a DVM; check the BC and BE junctions still act as diodes and measure the resistance between C and E.
Regards,
Jon
>I'll look more closely at the diodes and try to get some soft recovery types.
The existing MUR420 diodes are probably fine. Many people swear by the MUR860 which is just a higher voltage / current version. You could just add a small cap across each, perhaps in series with a resistor. To calculate the snubber correctly you'd need to know the inductance of the secondary winding of the transformer (which I don't) but you might start experimenting with something of the order of 2 to 20nF.
>From what you say - leaving the servo intact and using a polypropylene cap to block the DC - seems like the way to go.
That's probably what I'll do. Change C12 and C102 for something like 2.2uF or more (I think they are Rubycon in a grey sleeve at the moment). You could also change C8 / C108 which form part of the DC-servo but since it isn't directly in the audio path I probably wouldn't bother, besides you'd never get a polypropylene large enough (to avoid rolling off LF) - it's currently a 100uF and that's not practical.
>The deciding factor for replacing the 2 x 10000uF caps will be space. I'll have a look later today and see what might fit.
I don't think that they laid out the pcb for the wider lead pitch of larger caps , the 22000uF will probably have screw terminals anyway so you'll have to be creative in mounting them.
>How do you check the pre-drivers? Is there a test or do you just listen?
The circuit won't operate correctly with the output MOSFETs removed but you could look at the waveforms on a 'scope or just test the pre-drivers (standard NPN and PNP transistors) with a DVM; check the BC and BE junctions still act as diodes and measure the resistance between C and E.
Regards,
Jon
Hi Jon
Maybe I'll just wait for you to do yours - and then copy you
Would Nichicon Muse KZ/FG be okay for these positions? I have quite a few of them.
I had another look and there is some space between the 10000uF caps and the MOSFETS/Heatsink. There's also a few holes in the PCB around the large caps - but I don't know what is on the other side.
I'll do that. It will be good to know before I start taking it apart.
While the cover was off, I looked for the LM393 but couldn't find one. There are two pots on the pcb - one just behind the tone controls and one in the centre of the board. So, I'll probably need to get hold of a signal generator. Is it possible to apply the 1W rms 1KHz sine wave signal from a Test CD in a CD player?
Thanks
Fin
JonHarrison said:
Maybe I'll just wait for you to do yours - and then copy you
JonHarrison said:
Would Nichicon Muse KZ/FG be okay for these positions? I have quite a few of them.
JonHarrison said:
I had another look and there is some space between the 10000uF caps and the MOSFETS/Heatsink. There's also a few holes in the PCB around the large caps - but I don't know what is on the other side.
JonHarrison said:
I'll do that. It will be good to know before I start taking it apart.
While the cover was off, I looked for the LM393 but couldn't find one. There are two pots on the pcb - one just behind the tone controls and one in the centre of the board. So, I'll probably need to get hold of a signal generator. Is it possible to apply the 1W rms 1KHz sine wave signal from a Test CD in a CD player?
Thanks
Fin
Fin,
I have no experience of using the Nichicon Muse personally so I'm afraid I can't comment on them although I know people have replaced the FG with Elna Silmic and Cerafine with improved results.
>While the cover was off, I looked for the LM393 but couldn't find one.
Could you post a photo of the inside of the amp so that I can see what's what ?
>So, I'll probably need to get hold of a signal generator. Is it possible to apply the 1W rms 1KHz sine wave signal from a Test CD in a CD player?
1W / 8R requires 353mA RMS. I doubt a signal generator (and certainly not a CD player) could deliver that much current. You might just be better off setting to 5mV cold and then wait for 15 minutes until things warm up (and nothing smokes!) and feed a signal through the amp so that it's actually doing something and re-adjust for 4mV but I don't know what level the signal would have to be and whether it would really matter that much anyway!
Regards,
Jon
I have no experience of using the Nichicon Muse personally so I'm afraid I can't comment on them although I know people have replaced the FG with Elna Silmic and Cerafine with improved results.
>While the cover was off, I looked for the LM393 but couldn't find one.
Could you post a photo of the inside of the amp so that I can see what's what ?
>So, I'll probably need to get hold of a signal generator. Is it possible to apply the 1W rms 1KHz sine wave signal from a Test CD in a CD player?
1W / 8R requires 353mA RMS. I doubt a signal generator (and certainly not a CD player) could deliver that much current. You might just be better off setting to 5mV cold and then wait for 15 minutes until things warm up (and nothing smokes!) and feed a signal through the amp so that it's actually doing something and re-adjust for 4mV but I don't know what level the signal would have to be and whether it would really matter that much anyway!
Regards,
Jon
Hi Fin,
>What sort of equipment is generally used to provide the required signal? Maybe someone down here has what is needed.
I'm afraid I have no idea what is normally used but a signal generator fed through a power amp would do the trick. This is probably all over complicating things though. You may not need to adjust the bias at all or if you do, not by very much - the amp was setup properly before the output blew after all. Why don't you replace the mosfets and measure the voltage across the resistors first.
If you really want to feed in a test signal, as an alternative to a sig gen you could always burn a 1KHz test tone onto a CD.
Regards,
Jon
>What sort of equipment is generally used to provide the required signal? Maybe someone down here has what is needed.
I'm afraid I have no idea what is normally used but a signal generator fed through a power amp would do the trick. This is probably all over complicating things though. You may not need to adjust the bias at all or if you do, not by very much - the amp was setup properly before the output blew after all. Why don't you replace the mosfets and measure the voltage across the resistors first.
If you really want to feed in a test signal, as an alternative to a sig gen you could always burn a 1KHz test tone onto a CD.
Regards,
Jon
Hi Jon
I did some internet shopping yesterday and found some Nichicon LK 10,000uF caps going really cheap and bought eight of them. They are the same as those used by Arcam in one of their later amps and liked by the guys in this thread:http:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=41159&highlight=
Also managed to contact an Arcam service technician here in OZ - and he suggests that - as I don't have the correct equipment - I should measure the bias setting on the working channel, then install the new MOSFETS in the faulty channel and adjust to the same value as the good channel within one minute (to prevent the new MOSFETS from going bang). Then leave the amp switched on for about an hour with the speakers connected and the volume at min. Then readjust to match the good channel.
What do you think?
Regards
Fin
JonHarrison said:
I did some internet shopping yesterday and found some Nichicon LK 10,000uF caps going really cheap and bought eight of them. They are the same as those used by Arcam in one of their later amps and liked by the guys in this thread:http:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=41159&highlight=
Also managed to contact an Arcam service technician here in OZ - and he suggests that - as I don't have the correct equipment - I should measure the bias setting on the working channel, then install the new MOSFETS in the faulty channel and adjust to the same value as the good channel within one minute (to prevent the new MOSFETS from going bang). Then leave the amp switched on for about an hour with the speakers connected and the volume at min. Then readjust to match the good channel.
What do you think?
Regards
Fin
My friend's got an Arcam delta 290 with dead output transistors SM042A which need replacement but he found no information about this type of transistors.
Can anyone help to tell which equivalent transistor can be used for the instead?
Tks,
MK
Can anyone help to tell which equivalent transistor can be used for the instead?
An externally hosted image should be here but it was not working when we last tested it.
Tks,
MK
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.