Thanks for sharing with us, your findings Graeme. Very useful as I am in the process of adjusting my A-X's. I think I will solder in the 10k resistors now when the amps taken apart anyway. I wasnt quite aware that the absolute DC offset take so long time to settle. I was a bit put off to find 11V initially, thinking something was wrong, but I will try and let it cook for while longer.
Steen
Steen
Hi Ian,
I agree regarding the 30 ohms. It's just that NP uses 100 ohms and I just didn't feel like wasting the current. 30 ohms is fairly significant wrt 8 ohms.
I am a bit surprised that so few people have responded my last post. It's been almost four years since the Aleph-X thread started and there have been at least four different board sets for sale. That means there must be a large number of amplifiers running out there. To my knowledge only Grey and one or two others have built amps without these MacMillan resistors. You would think this would be a topic of interest. Perhaps I should have posted this particular discussion on the Aleph-X builders thread. What do you think?
There is a mistake in my post. I said below 30Hz regarding the improved bass. I should have said below 45Hz.
Hi Steen,
Good to hear from you. Let me know what you hear when you try the different resistor values.
I found that when I changed from 4.7K to either 10K or 22K I had to go throgh a complete re-adjustment cycle on the absolute DC offset. In fact after I removed the 4.7K resistors and had no resistors in place I turned on the amp and read a solid 25V output. I realize now that I should have tried to do an adjustment and listening cycle with no MacMillan resitors in order to establish a real baseline. Maybe I'll go back and try that. For now I'm having too much fun with the music.
Regards,
Graeme
I agree regarding the 30 ohms. It's just that NP uses 100 ohms and I just didn't feel like wasting the current. 30 ohms is fairly significant wrt 8 ohms.
I am a bit surprised that so few people have responded my last post. It's been almost four years since the Aleph-X thread started and there have been at least four different board sets for sale. That means there must be a large number of amplifiers running out there. To my knowledge only Grey and one or two others have built amps without these MacMillan resistors. You would think this would be a topic of interest. Perhaps I should have posted this particular discussion on the Aleph-X builders thread. What do you think?
There is a mistake in my post. I said below 30Hz regarding the improved bass. I should have said below 45Hz.
Hi Steen,
Good to hear from you. Let me know what you hear when you try the different resistor values.
I found that when I changed from 4.7K to either 10K or 22K I had to go throgh a complete re-adjustment cycle on the absolute DC offset. In fact after I removed the 4.7K resistors and had no resistors in place I turned on the amp and read a solid 25V output. I realize now that I should have tried to do an adjustment and listening cycle with no MacMillan resitors in order to establish a real baseline. Maybe I'll go back and try that. For now I'm having too much fun with the music.
Regards,
Graeme
Hi Graeme,
I also did some experimenting with the McMillan resistors but only went down. From 4k7 to 3k3, trying as low as 2k7.
When I did this I was mostly concentrating on absolute DC Offset and it´s behaviour over time and did no special listening tests with the different settings (I did listen but didn´t notice a difference)
At the moment the dc starts at 5-6V and goes down very rapidly to 1-2 volt and then to almost zero over the next hour.
I´ve used 78R for the output resistors (5 x 390R).
I will put in 10k in the near future and have a listen.
Thanks,
William
I also did some experimenting with the McMillan resistors but only went down. From 4k7 to 3k3, trying as low as 2k7.
When I did this I was mostly concentrating on absolute DC Offset and it´s behaviour over time and did no special listening tests with the different settings (I did listen but didn´t notice a difference)
At the moment the dc starts at 5-6V and goes down very rapidly to 1-2 volt and then to almost zero over the next hour.
I´ve used 78R for the output resistors (5 x 390R).
I will put in 10k in the near future and have a listen.
Thanks,
William
Hi Graeme,
I just did as William. Most concentrated on offset, and didn't perform listening tests Too bad. Was just living happy with my 4K75, but you raised some doubt... I won't sleep well the few next nights Till I make some tests...
But I'd like to understand how the IMMR (Ian McMillan Resistors) can be involved in the frequency response of the amp. Does anyone has a clue ? Is it a kind of bypass for the input capacitance of the input mosfets ? Assuming the whole amp is perfectly matched and symetrical, the IMMR nodal point is a virtual ground at AC.... Do I miss something ? Need to dig it a bit, but brother, can you spare a hint...
Among the other things I'd like to test is to try another input diff pair, such as Zetex mosfets (ZVP33xx). Really love these little thingies for their low capacitance. But need high currents to keep a low output Z of the first stage (low drain resistors, but 4V needed), so cascoding would be necessary here to lower dissipation. Has anyone experimented ?
So many things, so little time...
I just did as William. Most concentrated on offset, and didn't perform listening tests
Too bad. Was just living happy with my 4K75, but you raised some doubt... I won't sleep well the few next nights Till I make some tests...But I'd like to understand how the IMMR (Ian McMillan Resistors
) can be involved in the frequency response of the amp. Does anyone has a clue ? Is it a kind of bypass for the input capacitance of the input mosfets ? Assuming the whole amp is perfectly matched and symetrical, the IMMR nodal point is a virtual ground at AC.... Do I miss something ? Need to dig it a bit, but brother, can you spare a hint...Among the other things I'd like to test is to try another input diff pair, such as Zetex mosfets (ZVP33xx). Really love these little thingies for their low capacitance. But need high currents to keep a low output Z of the first stage (low drain resistors, but 4V needed), so cascoding would be necessary here to lower dissipation. Has anyone experimented ?
So many things, so little time...
Thank you to everyone for posting. I have another correction. The key comments by NP including the "going to hell" one is on page 34 of the Aleph-X page. It seems to me that NP was trying to tell people (loudly) to use moderation with regard to the IMMR's but everyone was fixating on DC drift.
William,
I look forward to hearing what you think. I have always appreciated the advice and help you've given others here.
Cheff,
I didn't mean to disturb your sleep. The input is a virtual ground. The output node is very low impedance. The diff pair sources by comparison are a high impedance node set by the CCS. The IMMR's can be seen as being in parallel with this impedance and can lower it quite a bit. IMO whatever process is suppressing the absolute DC drift can reach up into the low bass region if too mach feedback is applied from the output. But I'm just speculating. And if you do any experiments please post the results.
I have a bag of ZVP3310's as well. Very nice parts. Even 6 volts at 10ma is only 60mw. That's fine for a mosfet that's rated for 600mw. Then use your existing 9610's as the cascode transistors. Set the cascode voltage with a 9V zener and you're in business. Let me know how it turns out. You really didn't need to sleep did you?
Algar_emi,
It's a Signal 88-8 which has an 88VAC secondary at 8A. So I'm figuring 704VA plus 10% because of the 108V and 128V primary taps. My guess would be a conservative 750VA.
Regards,
Graeme
William,
I look forward to hearing what you think. I have always appreciated the advice and help you've given others here.
Cheff,
I didn't mean to disturb your sleep. The input is a virtual ground. The output node is very low impedance. The diff pair sources by comparison are a high impedance node set by the CCS. The IMMR's can be seen as being in parallel with this impedance and can lower it quite a bit. IMO whatever process is suppressing the absolute DC drift can reach up into the low bass region if too mach feedback is applied from the output. But I'm just speculating. And if you do any experiments please post the results.
I have a bag of ZVP3310's as well. Very nice parts. Even 6 volts at 10ma is only 60mw. That's fine for a mosfet that's rated for 600mw. Then use your existing 9610's as the cascode transistors. Set the cascode voltage with a 9V zener and you're in business. Let me know how it turns out. You really didn't need to sleep did you?
Algar_emi,
It's a Signal 88-8 which has an 88VAC secondary at 8A. So I'm figuring 704VA plus 10% because of the 108V and 128V primary taps. My guess would be a conservative 750VA.
Regards,
Graeme
CheffDeGaar said:Hi Graeme,
I'd like to understand how the IMMR (Ian McMillan Resistors
Where can a guy get more information regarding MacMillan resistors? Where can they be found?
Thanks!
-David
I have a few questions about the startup/adjustment procedure. Hope you can help out. When measuring the voltage over the source resistors in my A-X's, I get a reading of about 0,6V initially but the value rises. When it get close to 1V I get concerned and turn off the amp. The source resistors are 0,47r so I guess I should be able to adjust for 0,47V to have 1 amp flowing through each fet? I have 12 fet's per monoblock, and 1000VA, 2x18V trafo secondaries which gives just over +/-24V at no load, when the amp is connected to the supply, the rails drops to about 19V, I think thats a lot of drop?. I increased the value of R14/R31 and R15/R32 to 1,5k and kept the trimpots VR1 and VR2 (set to minimum resistance). Will the voltage drop over the source resistors settle down over time as the absolute DC does?
There is, as I wrote in the post above approx 11V absolute DC offset and there is next to no relative DC at the outputs. Seemingly the amp works and the heatsinks holding the outputdevices works too, thats for sure. The voltage drop over the source resistors are the same, both at the output devices and the CCS devices.
Gate to Source voltage is Approx. 4,6V on all the devices.
Should I be concerned about the rising source voltage (as I am) or should I let the amp heat up for a longer period (which I didnt risk at this point)?
Steen
There is, as I wrote in the post above approx 11V absolute DC offset and there is next to no relative DC at the outputs. Seemingly the amp works and the heatsinks holding the outputdevices works too, thats for sure. The voltage drop over the source resistors are the same, both at the output devices and the CCS devices.
Gate to Source voltage is Approx. 4,6V on all the devices.
Should I be concerned about the rising source voltage (as I am) or should I let the amp heat up for a longer period (which I didnt risk at this point)?
Steen
gl said:
Hi Graeme,
Well, I did have a very good night, in fact
WRT IMMRs and impedance, I sure share your opinions... An influence of a parasitic capacitance somewhere ? Will perform some tests, but this won't be right now...
I do not plan to use a mosfet as a cascode element (sorry
). I'd like to get a decent Vds on the input mosfets, just to keep the capacitance as low as possible. I'm aiming at about 10V on the mosfets (so about 100mW), and use a BJT for cascoding, so I won't "loose" 4V in the cascode. Need enough swing on the drain resistors voltage (theoretically 20mA x Rd, so about 8V), and as my rails are @ 21V, fits perfectly for a bjt here Any particular reason for using a mosfet here ?With a friend of mine, we're designing a pcb for the AX trying to allow for all this kind of modifications : Various current sources for the input pair, plugable input diff pair (IRF, ZVP, 2SJ, etc...) and so on, just to have an easy beast of burden for all our tests
. And of course, we will report our findings here.Cheers
Steenoe,
maybe you should lower the resistors that are in series with VR1 and VR3. I changed them to 27k.
19V seems very low! Normally transformers are specified at rated power. With big torroids this means 3-5% more voltage at no load. Do you use a C-R-C supply?
Normally the voltage over the source resistors goes down with temperature not up!
William
maybe you should lower the resistors that are in series with VR1 and VR3. I changed them to 27k.
19V seems very low! Normally transformers are specified at rated power. With big torroids this means 3-5% more voltage at no load. Do you use a C-R-C supply?
Normally the voltage over the source resistors goes down with temperature not up!
William
Hi William. Thanks for the reply. I am using a C-L-C supply consisting of 68.000uF-2mH-68.000uF on each rail. I had the suspicion that too much current is running through the amp.
It just came to mind that the voltage drop over the output to GND resistor is around 12V, that seems like a lot to me also. They are at 124r as far as I remember.
I did lower the resistors in series with the pot's but that helped only a little (I have around 27K there too), and the voltage keep rising. I have triple checked all connections, so I dont excpect any faults, but I better check again.
Steen
It just came to mind that the voltage drop over the output to GND resistor is around 12V, that seems like a lot to me also. They are at 124r as far as I remember.
I did lower the resistors in series with the pot's but that helped only a little (I have around 27K there too), and the voltage keep rising. I have triple checked all connections, so I dont excpect any faults, but I better check again.
Steen
Hi Dave,
The only advice I can give you about info on the MacMillan resistors is to read the first 40 or 50 pages of the big Aleph-X thread. It's all there in bits and pieces.
Hi Steen,
I agree with William. On my amps the Source resistor voltage drops 25mv from turn-on to full operating temperature on the heat sinks. Your absolute DC offset should not be staying at 11V. You should initially adjust it to as close to zero as possible over several hours (or days). From then on it should settle down within an hour after turn-on to something you can live with - for me that's less than a volt. Check the voltage across the BJT control transistor in the Aleph CS. It should be 4 to 5 volts. I suspect there is a wiring error in your amp or a part with the wrong value. It's just a guess. (BTW with the first channel I turned on I discovered I had all the main rectifiers installed backwards. I also had a serious error in the Base circuit of the CCS BJT. And I had checked all the connections three times!)
Hi Cheff,
Yes a BJT would make the best cascode and for all the reasons you state. I just suggested the mosfets because if you already have a working amp with 9610's as the diff pair it would be easy and cheap to try the mod you described.
I'm not sure that the ZVP33xx is going to make a big enough difference to affect the sound of the amp. My preferred solution would be to use a 2SJ109 with BJT cascodes. I know that this has been discussed many times and I am aware of the current drive/power dissipation limitations, but I am currently building a new version of this amp based on this idea.
I look forward to hearing about the results of your work with your new board.
Cheers,
Graeme
The only advice I can give you about info on the MacMillan resistors is to read the first 40 or 50 pages of the big Aleph-X thread. It's all there in bits and pieces.
Hi Steen,
I agree with William. On my amps the Source resistor voltage drops 25mv from turn-on to full operating temperature on the heat sinks. Your absolute DC offset should not be staying at 11V. You should initially adjust it to as close to zero as possible over several hours (or days). From then on it should settle down within an hour after turn-on to something you can live with - for me that's less than a volt. Check the voltage across the BJT control transistor in the Aleph CS. It should be 4 to 5 volts. I suspect there is a wiring error in your amp or a part with the wrong value. It's just a guess. (BTW with the first channel I turned on I discovered I had all the main rectifiers installed backwards. I also had a serious error in the Base circuit of the CCS BJT. And I had checked all the connections three times!)
Hi Cheff,
Yes a BJT would make the best cascode and for all the reasons you state. I just suggested the mosfets because if you already have a working amp with 9610's as the diff pair it would be easy and cheap to try the mod you described.
I'm not sure that the ZVP33xx is going to make a big enough difference to affect the sound of the amp. My preferred solution would be to use a 2SJ109 with BJT cascodes. I know that this has been discussed many times and I am aware of the current drive/power dissipation limitations, but I am currently building a new version of this amp based on this idea.
I look forward to hearing about the results of your work with your new board.
Cheers,
Graeme
Hi Graeme,
I replaced the 3k3 McMillan (R46/47) resistors with 10k ones yesterday.
Absolute dc behaviour changed from 6V at startup to 16V while it takes a lot longer to come below 2V. I´m not shure I like this and will try some other values too.
I did listen a bit yesterday but since I´ve changed quite a bit the last few months (cascoded x-bosoz, cd-transport, room) it is not so easy to tell the difference.
First impressions:
-no change in the lower frequencies
-the mid/highs did change (and this sounds a bit funny) sounding more free and powerfull and at the same time more "rounded". On some CD´s this was almost too much (the power I mean)
I will have to listen a bit more and go back to 6k8 or 4k7 to be shure I don´t imagine things.......
William
I replaced the 3k3 McMillan (R46/47) resistors with 10k ones yesterday.
Absolute dc behaviour changed from 6V at startup to 16V while it takes a lot longer to come below 2V. I´m not shure I like this and will try some other values too.
I did listen a bit yesterday but since I´ve changed quite a bit the last few months (cascoded x-bosoz, cd-transport, room) it is not so easy to tell the difference.
First impressions:
-no change in the lower frequencies
-the mid/highs did change (and this sounds a bit funny) sounding more free and powerfull and at the same time more "rounded". On some CD´s this was almost too much (the power I mean)
I will have to listen a bit more and go back to 6k8 or 4k7 to be shure I don´t imagine things.......
William
Hi,
10k probably is a bit too high for my amps. Voltage keeps creeping down even after 3 or 4 hours of playing. Startup would be something like 18-20V wich is almost rail voltage if I would compensate for this.
So now I put 6k8 in one channel and will see what happens.
If I remember correctly when I first build the amps with zeners as a voltage reference one amp went from plus to minus and the other one was the other way round. Changing the zeners for precision voltage references made both amps behave the same way (going from plus to minus).
Wouldn´t it be possible to look for a zener that sits between the two I had in my amps and lets the amp stay at 0V from startup? Or some other temperature compensating part?
William
10k probably is a bit too high for my amps. Voltage keeps creeping down even after 3 or 4 hours of playing. Startup would be something like 18-20V wich is almost rail voltage if I would compensate for this.
So now I put 6k8 in one channel and will see what happens.
If I remember correctly when I first build the amps with zeners as a voltage reference one amp went from plus to minus and the other one was the other way round. Changing the zeners for precision voltage references made both amps behave the same way (going from plus to minus).
Wouldn´t it be possible to look for a zener that sits between the two I had in my amps and lets the amp stay at 0V from startup? Or some other temperature compensating part?
William
Hi Algar_emi,
Your transformer will yield 30 * 1.414 = 42V (approximately) for power rails. This is too high, even if you used R-C or L-C sections to drop the voltage. Besides you would waste too much power. A transformer with 18V to 24V secondaries would be in the ballpark.
Hi William,
Generally speaking you seem to be getting similar results to mine as far as the quality of the sound goes. I understand your situation about not wanting to change too many things at once. At this point I am getting used to turning the amps on 30 to 60 minutes before starting to play music. The PassLabs manuals recommend doing this. Maybe this is part of the reason. I think that Nelson Pass is much more tolerant of absolute offset than most of the DIYers here. I am teaching myself to be braver about this because I believe that there are sonic benefits.
I'm afraid I don't remember any details about your Aleph-X and I didn't understand the details of your voltage reference remarks. Is there a thread reference you could give me? Both of my amps go from plus to minus.
Cheers,
Graeme
Your transformer will yield 30 * 1.414 = 42V (approximately) for power rails. This is too high, even if you used R-C or L-C sections to drop the voltage. Besides you would waste too much power. A transformer with 18V to 24V secondaries would be in the ballpark.
Hi William,
Generally speaking you seem to be getting similar results to mine as far as the quality of the sound goes. I understand your situation about not wanting to change too many things at once. At this point I am getting used to turning the amps on 30 to 60 minutes before starting to play music. The PassLabs manuals recommend doing this. Maybe this is part of the reason. I think that Nelson Pass is much more tolerant of absolute offset than most of the DIYers here. I am teaching myself to be braver about this because I believe that there are sonic benefits.
I'm afraid I don't remember any details about your Aleph-X and I didn't understand the details of your voltage reference remarks. Is there a thread reference you could give me? Both of my amps go from plus to minus.
Cheers,
Graeme
Hi Graeme,
here´s the thread:
http://www.diyaudio.com/forums/showthread.php?threadid=49528
I would like to measure a "standard" AX-160 to see how it behaves. I don´t think it´ll start up at 16-20V though.
William
P.S. the 6k8 seems to be OK. Startup is 12V and it stays around 0V after a few hours. I listened to it while the other channel still had the 10k resistors. Didn´t hear any unbalance........
I will put the 6k8 in the other channel today and listen a bit more.
William
here´s the thread:
http://www.diyaudio.com/forums/showthread.php?threadid=49528
I would like to measure a "standard" AX-160 to see how it behaves. I don´t think it´ll start up at 16-20V though.
William
P.S. the 6k8 seems to be OK. Startup is 12V and it stays around 0V after a few hours. I listened to it while the other channel still had the 10k resistors. Didn´t hear any unbalance........
I will put the 6k8 in the other channel today and listen a bit more.
William
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.