Argueing over what all this means...... The answer to that is easy... the building of more amps
zvp3310a
Bob,
I eventually, replaced all three with the zvp3310a. Everything seems to be fine. I have not had any problems. Yes, they have to be matched.
I purchased some to match. The first time the DC offset was way too high with over 100 mV dc offset. The second set I tried a different current value and it was much better around 40 to 50 mV. Wth the original irf9610 the zvp3310 differential were much closer around 10 to 15 mV. For some reason they seemed to go worst with all zvp3310a in the ciruit.
I noticed when I initially left the irf9610 for the current source, the two matched zvp3310 were a lot closer matched on the dc offset. When I added the 3rd zvp3310a into the current source for both of them, I found the dc offset drifted higher.
Does anyone know how to measure the Aleph 3 for how much current the current source is drawing and also to check the differential for current draw?
I would like to rematch some down the road, but would like to match them for the current load of my specific aleph rather than just assume 10 mA for each differential and 20 ma for the current source.
Bill,
Bob,
I eventually, replaced all three with the zvp3310a. Everything seems to be fine. I have not had any problems. Yes, they have to be matched.
I purchased some to match. The first time the DC offset was way too high with over 100 mV dc offset. The second set I tried a different current value and it was much better around 40 to 50 mV. Wth the original irf9610 the zvp3310 differential were much closer around 10 to 15 mV. For some reason they seemed to go worst with all zvp3310a in the ciruit.
I noticed when I initially left the irf9610 for the current source, the two matched zvp3310 were a lot closer matched on the dc offset. When I added the 3rd zvp3310a into the current source for both of them, I found the dc offset drifted higher.
Does anyone know how to measure the Aleph 3 for how much current the current source is drawing and also to check the differential for current draw?
I would like to rematch some down the road, but would like to match them for the current load of my specific aleph rather than just assume 10 mA for each differential and 20 ma for the current source.
Bill,
Blues suggests a good solution . . .
By the way, I would simply measure the voltage across R108 and divide it with R108 to get the bias current of Q101.
Then, I would measure of Vgs of Q101 and choose Q102 having exactly the same Vgs as Q101 . . . If this is difficult . . . I would try to adjust R106 for Q101 and Q102 to have the same Vgs . . .
jh
By the way, I would simply measure the voltage across R108 and divide it with R108 to get the bias current of Q101.
Then, I would measure of Vgs of Q101 and choose Q102 having exactly the same Vgs as Q101 . . . If this is difficult . . . I would try to adjust R106 for Q101 and Q102 to have the same Vgs . . .
jh
Matching zvp3310a for tighter DC offset.
Blues,
What you are describing sounds like adjusting the AC current gain to the ideal 50 percent?
I already have it adjusted with pots already to that 50 percent value.
I want to rematch my zvp3310 stock pile to tighter dc offset. I am trying to find their actual current draw for the differential and current source to them so that I can rematch say 90 plus zvp3310's.
Are you saying the input side is similar to the output's AC current gain?
I do not have the Aleph schematic near by so would have to look those up later.
Regards, Bill
Blues,
What you are describing sounds like adjusting the AC current gain to the ideal 50 percent?
I already have it adjusted with pots already to that 50 percent value.
I want to rematch my zvp3310 stock pile to tighter dc offset. I am trying to find their actual current draw for the differential and current source to them so that I can rematch say 90 plus zvp3310's.
Are you saying the input side is similar to the output's AC current gain?
I do not have the Aleph schematic near by so would have to look those up later.
Regards, Bill
Bill,
While matching in the circuit is great it is also very time consumming. Why not just match them the Nelson Pass way? Build a small test jig with a regulated supply and measure the voltage drop of each device?? I'f loan you my 51/2 digit DMM for that chore. Its alot quicker and just as accurate in getting a very low DC offset. A search turned up this nifty matching circuit .
If going this route doesn't get you low dc offset then something else is amuck.
Mark
While matching in the circuit is great it is also very time consumming. Why not just match them the Nelson Pass way? Build a small test jig with a regulated supply and measure the voltage drop of each device?? I'f loan you my 51/2 digit DMM for that chore. Its alot quicker and just as accurate in getting a very low DC offset. A search turned up this nifty matching circuit .
If going this route doesn't get you low dc offset then something else is amuck.
Mark
Re: zvp3310a
BillWW said:Bob,
I eventually, replaced all three with the zvp3310a. Everything seems to be fine. I have not had any problems. Yes, they have to be matched.
I purchased some to match. The first time the DC offset was way too high with over 100 mV dc offset. The second set I tried a different current value and it was much better around 40 to 50 mV. Wth the original irf9610 the zvp3310 differential were much closer around 10 to 15 mV. For some reason they seemed to go worst with all zvp3310a in the ciruit.
I noticed when I initially left the irf9610 for the current source, the two matched zvp3310 were a lot closer matched on the dc offset. When I added the 3rd zvp3310a into the current source for both of them, I found the dc offset drifted higher.
Hi Bill,
Thanks a lot for this clarification, here.
It is interesting that the apparent 'sensitivity' re. DC offsets changed with different current source fets. Did you notice any difference 'sonically' when you later swapped out the current source fet to a Zetex? Otherwise, I wonder why perhaps you didn't return to the IRF component for easier DC-setting characteristics.
IIRC, Nelson suggested (at least in some later Alephs) that these LTPs ideally need to be 'matched in circuit' for lowest overall distortion, and then take care of any DC offsets by tweeking a couple of the resistors' values. It sounds like this might be easier to do with retaining the original IRFs for the current source, unless there is a sonic improvement to using the Zetex fets here.
Anyway, congrats on what seems to be a worthwhile sonic improvement here.
Regards,
Other Zetex Mosfets
I have a mini board laying around in my pile. Going to try ZVP3306 for the input pair and ZVP2106 for the current source. This higher current small mosfet may help keep the offsets down.
My speakers are easier damaged than most by large amounts of dc. I always worried about more than single digit mv levels. While wasting several months playing with Tripath amps the levels varied all over the place. Some were in the 100 mv range.
My speakers are crossoverless, 13 ohm dcr, and wound with about 36 ga wire. Being powered up for days with this 100 mv of dc did nothing to hurt the drivers.
In fact I expected to see offset in driver position, and be able to here the channels with the higher offsets. No such luck, sometimes I had a 100 mv on one channel and 30 on the other. Sonically they were the same.
But reason came back, the Tripath are in with the other toys and reals amps with less than 5 mv of offset have taken their place.
I hope my results from trying these small mosfets in place of the IRF9610's mirror is as positive.
Btw, to match up your mosfets, measure the resistance from source to drain (or drain to source here) on the lowest ohm setting. It usually drops for about a minute, then stabilizes. In my junk, this method has given current draws in the 10% range. With is most likely close to what you will get after heating and drift effects from better matched parts.
George
I have a mini board laying around in my pile. Going to try ZVP3306 for the input pair and ZVP2106 for the current source. This higher current small mosfet may help keep the offsets down.
My speakers are easier damaged than most by large amounts of dc. I always worried about more than single digit mv levels. While wasting several months playing with Tripath amps the levels varied all over the place. Some were in the 100 mv range.
My speakers are crossoverless, 13 ohm dcr, and wound with about 36 ga wire. Being powered up for days with this 100 mv of dc did nothing to hurt the drivers.
In fact I expected to see offset in driver position, and be able to here the channels with the higher offsets. No such luck, sometimes I had a 100 mv on one channel and 30 on the other. Sonically they were the same.
But reason came back, the Tripath are in with the other toys and reals amps with less than 5 mv of offset have taken their place.
I hope my results from trying these small mosfets in place of the IRF9610's mirror is as positive.
Btw, to match up your mosfets, measure the resistance from source to drain (or drain to source here) on the lowest ohm setting. It usually drops for about a minute, then stabilizes. In my junk, this method has given current draws in the 10% range. With is most likely close to what you will get after heating and drift effects from better matched parts.
George
zvp3310a for current source.
The main thing that seemed to occur in using the zvp3310a's are that they drift more than the larger irf9610.
When I turn the amp on, the dc offset is around 3 or 4 mV on the zvp3310a and then it drifts gradually up to 40 mV on one channel and at 57 degrees Celsius on the output heatsinks, the right channel drifts to about 48 mV and they both remain there plus or minus a few mV.
It seemed like the irf9610 drifted about half the values of my zvp3310a. Keep in mind, I have only tested two sets of zvp3310a on my left and right channels. I have not done a lot so it could be just my zvp3310a batch.
But, it still seems to be occuring because the 9610 has a larger die, which keeps them more linear. The 9610's die acts as a heatsink, which seems to stabilize those larger to-220 mosfets versus the plastic to-92 zvp3310a casing.
Mark suggested perhaps adding a copper wire wrapped around the differential with heat sink compound, when I was at his home. He said perhaps someone could add some epoxy around the two differentials to keep them more equalized too.
Regarding sound quality, I personally found the zvp3310a for a current source to sound more musical than the irf9610, so I decided to leave it in. Maybe knowing the irf9610 has 4 times the input capacitance value compared to the zvp3310a part influenced my ears? It seemed to open up the details a little bit more and a little bit more spaciousness I thought at the time, so I just left it in. The irf9610 sounded a little dry to my ears. The zvp3310 sounded sweeter in my listening sessions.
Thanks guys for the assistance on matching. I was using a regulated 15.00 volt supply that seemed to be pretty stable. I will have to see what Mark is talking about on his power supply to see if mine is good enough. Maybe again Saturday I could swing by, since I have a lot more fets to match, if his way is easier than mine.
The zvp3310a need their first two pins switched around to match the irf9610 pin outs.
The ZVP3310 are a direct drop in replacement, so it is an uncomplicated mod to perform. Most people should be able to do this relatively easy for a great improvement in sound quality. I do not think only the zvp3310 needs to be used. I think anything with low capacitance and enough current rating to handle 10 or 20 mA would be fine. If you find something lower on the input capacitance ,try it out. Find something lower than 50 pf and let us know your results! It really does change the quality of the aleph a great deal!!!
I did have to readjust the AC current gain after placing them in the circuit, which I always check when performing mods. I have found the amp can sound quite dull in the upper frequencies if this is not checked making it sound not too impressive.
I think aside from adding various parts, the AC current gain needs to be at the ideal 50 percent, which seemed to also bring out the details and highs in recordings. I use a 10 turn pot to adjust this parameter, which makes it easier to adjust.
Bill,
The main thing that seemed to occur in using the zvp3310a's are that they drift more than the larger irf9610.
When I turn the amp on, the dc offset is around 3 or 4 mV on the zvp3310a and then it drifts gradually up to 40 mV on one channel and at 57 degrees Celsius on the output heatsinks, the right channel drifts to about 48 mV and they both remain there plus or minus a few mV.
It seemed like the irf9610 drifted about half the values of my zvp3310a. Keep in mind, I have only tested two sets of zvp3310a on my left and right channels. I have not done a lot so it could be just my zvp3310a batch.
But, it still seems to be occuring because the 9610 has a larger die, which keeps them more linear. The 9610's die acts as a heatsink, which seems to stabilize those larger to-220 mosfets versus the plastic to-92 zvp3310a casing.
Mark suggested perhaps adding a copper wire wrapped around the differential with heat sink compound, when I was at his home. He said perhaps someone could add some epoxy around the two differentials to keep them more equalized too.
Regarding sound quality, I personally found the zvp3310a for a current source to sound more musical than the irf9610, so I decided to leave it in. Maybe knowing the irf9610 has 4 times the input capacitance value compared to the zvp3310a part influenced my ears? It seemed to open up the details a little bit more and a little bit more spaciousness I thought at the time, so I just left it in. The irf9610 sounded a little dry to my ears. The zvp3310 sounded sweeter in my listening sessions.
Thanks guys for the assistance on matching. I was using a regulated 15.00 volt supply that seemed to be pretty stable. I will have to see what Mark is talking about on his power supply to see if mine is good enough. Maybe again Saturday I could swing by, since I have a lot more fets to match, if his way is easier than mine.
The zvp3310a need their first two pins switched around to match the irf9610 pin outs.
The ZVP3310 are a direct drop in replacement, so it is an uncomplicated mod to perform. Most people should be able to do this relatively easy for a great improvement in sound quality. I do not think only the zvp3310 needs to be used. I think anything with low capacitance and enough current rating to handle 10 or 20 mA would be fine. If you find something lower on the input capacitance ,try it out. Find something lower than 50 pf and let us know your results! It really does change the quality of the aleph a great deal!!!
I did have to readjust the AC current gain after placing them in the circuit, which I always check when performing mods. I have found the amp can sound quite dull in the upper frequencies if this is not checked making it sound not too impressive.
I think aside from adding various parts, the AC current gain needs to be at the ideal 50 percent, which seemed to also bring out the details and highs in recordings. I use a 10 turn pot to adjust this parameter, which makes it easier to adjust.
Bill,
Many thanks, Bill.
From what you now say, it does appear to be advantageous to replace all 3 fets here, and sort out any DC problems by other means. Certainly, some heatsinking on the LTPs is likely to reduce any DC drift, rather than increase, it in my experience.
Are your 2 LTP fets here 'tied together' in any way, so that any temperature effects are similar for both devices?
In similar situations in other amp designs, I have either used dual fets, or cemented them together (back-to-back) with sometimes a piece of aluminium sheet between them to act as a small sink.
I am sure that for most of us, any potential sonic improvement is valuable, even if it causes a few headaches elsewhere.
It is the *sound* which rules, after all!
Regards,
From what you now say, it does appear to be advantageous to replace all 3 fets here, and sort out any DC problems by other means. Certainly, some heatsinking on the LTPs is likely to reduce any DC drift, rather than increase, it in my experience.
Are your 2 LTP fets here 'tied together' in any way, so that any temperature effects are similar for both devices?
In similar situations in other amp designs, I have either used dual fets, or cemented them together (back-to-back) with sometimes a piece of aluminium sheet between them to act as a small sink.
I am sure that for most of us, any potential sonic improvement is valuable, even if it causes a few headaches elsewhere.
It is the *sound* which rules, after all!
Regards,
In the opening post of this thread, Mark says that the Krell KSA50 was the warmest sounding amp, whereas Bill seems to praise the modified Alephs.
If replacing the current source also to the ZVP made the Aleph more accurate, detailed and sweeter too, then does a lesser detailed 'warmer' sound have more appeal? And does the modified Aleph lack that "warmth" due to a more open and detailed midrange and top-end?
What are the opinions of others at the shootout??
If replacing the current source also to the ZVP made the Aleph more accurate, detailed and sweeter too, then does a lesser detailed 'warmer' sound have more appeal? And does the modified Aleph lack that "warmth" due to a more open and detailed midrange and top-end?
What are the opinions of others at the shootout??
The best way to describe it is that the Krell is like being within the first ten rows of a performance while with the Aleph you are about 20 rows further back... while detail is actually a little better. Ambience retrevial from the Aleph was a little better but not by much. The KSA-50 always did have great soundstaging abilities. That is the best way I can describe it. I think the Aleph modded will sound better on a horn loaded system since it would introduce less "honk" into the midrange. I think Luckylyndy can also confirm this.
I plan on modding my Mini Aleph as soon as I have time to unearth my ZVP's. They're somewhere around here.....
Mark
I plan on modding my Mini Aleph as soon as I have time to unearth my ZVP's. They're somewhere around here.....
Mark
Bobken said:Many thanks, Bill.
From what you now say, it does appear to be advantageous to replace all 3 fets here, and sort out any DC problems by other means. Certainly, some heatsinking on the LTPs is likely to reduce any DC drift, rather than increase, it in my experience.
Are your 2 LTP fets here 'tied together' in any way, so that any temperature effects are similar for both devices?
In similar situations in other amp designs, I have either used dual fets, or cemented them together (back-to-back) with sometimes a piece of aluminium sheet between them to act as a small sink.
I am sure that for most of us, any potential sonic improvement is valuable, even if it causes a few headaches elsewhere.
It is the *sound* which rules, after all!
Regards,
Bob,
My mikeW boards do not have the mosfets back to back. I really did not have too many problems in spite of this. My dc offset is stable between 40 to 46 mV offset, any ways, so it is far lower than 100 mV so I am not worrying about it.
I will try some of the ideas of matching for my circuit later, when I build another amp in the next couple weeks. I have tweaked this current amplifier so much that I am going to leave it alone, because a lot of the traces have ccome off from desoldering. I am just going to leave it alone for now. I might like to double the heatsinks and make it fanless later.
This amp was really just thrown together to use as a prototype to learn what to use in future Alephs.
Samuel Jayaraj said:
Samuel,
The Krell really sounded warm, because it seems to emphasize the bass and midrange more than the Alephs do.
The Aleph seemed to emphasize the highs more than its bass. It seemed to be balanced throughout very evenly with a softer sounding bass compared to the Krell.
I found the Krell was very punchy with its bass, although I still think the Aleph seemed to extend lower on the lowest bass notes more naturally. The Krell sounded more powerful above this lower frequency that the Aleph emphasized, which perhaps made the Krell seem to lose the lowest octave in the lower frequency range, while emphasing the upper bass to midrange well.
Mark is correct. I do not think the Krell would sound as well on my horn, because horn speakers tend to not need a lot of dampening factor or control like the Krell does.
I have tried a KST-100 on my horns and it still sounded okay, but the Aleph was more relaxing to listen to. I have also noticed with Krells in general that the lowest octives seem to be missing. I do not know why it is with the Krells I have heard. Most people would not notice this though. Mark and Levenson's that I have borrowed seemed to extend lower in the lowest octives too compared to the Krell KSA-150 that I have also compared it to.
Some people, might prefer a more warm sounding amp too. It just comes down to personal preference.
Build both of them and try them for yourself?
I am sure I could also tweak the Krell and it could sound better too. Tweaks do make a difference.
BTW, I do have a set of boards for the Krell too, so that will be another project to play with. I would like to use the Krell on some Apogee speakers that I own some day when I have a large enough room for those speakers. The aleph does not have enough control for those speakers, except for the bass panel, which seems easier to drive.
Regards, Bill
Correction on components used in Aleph 3
I realized I had posted 18 gage wire for the iron laminate inductors, but they are actually 15 gage, a much heavier gage magnet wire than what I posted.
Also, I referred to crossover nonpolar capacitors and then referred back to "electrolytics," which I should have said polar electrolytics.
Regards, Bill
I realized I had posted 18 gage wire for the iron laminate inductors, but they are actually 15 gage, a much heavier gage magnet wire than what I posted.
Also, I referred to crossover nonpolar capacitors and then referred back to "electrolytics," which I should have said polar electrolytics.
Regards, Bill
BillWW said:
Hi Bill,
I know just what you mean here with the traces lifting, and it is one of the hazards of this kind of exercise, caused by the enforced re-heating of the areas. There is no alternative, though, as you have found, if you do this kind of 'development' work.
If It helps, I have experienced some good results from cleaning up the boards and copper traces with flux-remover, and then sticking them back down again with super-glue (Cyano).
Just run some cyano under the traces where they are lifted, and I find a wooden coctail-stick soaked in (low-viscosity) cyano till a small 'blob' forms on the end, is good to position this glue in the right location, and capillary action will draw it between the parts. Then subsequently a bit of pressure on the trace, also with the point of the cocktail-stick pressing it down on the board, will set the cyano off, by excluding the air, as, of course, it is anaerobic in its method of setting.
Provided you get rid of all traces of flux etc. beforehand, this will give a reasonably durable solution, even if you need to do some more soldering, and the added 'support' to the flimsy copper traces provided by the board when this is done, is well worthwhile. I have several areas like this which are still good maybe 20 yrs later.
Considering this amp of yours was "just thrown together", you seem to have achieved a very good result, from all accounts.
Well done for an initial effort!
Regards,
Disecting the amps in additional detail.
Keep in mind to the sound quality for each channel of the amps.
I was using 262,000 uF vs the Krell of Marks was 112,000 uF.
I was using a 2.5 mH pi filter on top of my hi capacitance, while I do not believe Mark had any crc or pi network.
He was using 3 mosfets per output, while the Aleph 3 was using 2. The additional output mosfets of the Krell would add additional capacitance, which would roll the highs off.
The stock Krell 50 watt clone used 2, so it in theory should have higher frequency extension than 3 or more output mosfets in parallel.
The Krell uses 1.9 Amps of bias, while off the top of my head, I do not recall what mine is, but I am sure it was more like 2.3 or 2.5 amps of bias, so that could explain the additional spaciousness. My power line I measured with an AC RMS meter and it was drawing 4.7 Amps continuously at roughly 117 volts mains for my Aleph 3. Someone want to do the math to how much bias am I pushing? Is that 275 watts per channel minus transformer losses etc? I would guess 250 watts per channel for an Aleph 3 is quite a bit.
Hopefully, it will last with the fan cooling at 55 to 60 C.
Usually more bias keeps the mosfets more linear, which sounds better. So per ratio, his 50 Watt amp is using less bias per ratio compared to my 30 Watt Aleph.
The aleph would be a second harmonic amp, while the Krells push pull would be 3rd order harmonics? Two styles of sound.
I have tried my Aleph bridged to mono, which cancels the second harmonics, which I have grown fond of. It also rolled the highs off noticeably with back then essentially 4 mosfets paralleled.
Btw, there really was not that many people at this shoot out, so that is why you are not hearing very many opinions. My brother has not responded, because he is on a month long business trip.
I don't know about Lyndon though. He would be the only other person there to comment.
Regards, Bill
Keep in mind to the sound quality for each channel of the amps.
I was using 262,000 uF vs the Krell of Marks was 112,000 uF.
I was using a 2.5 mH pi filter on top of my hi capacitance, while I do not believe Mark had any crc or pi network.
He was using 3 mosfets per output, while the Aleph 3 was using 2. The additional output mosfets of the Krell would add additional capacitance, which would roll the highs off.
The stock Krell 50 watt clone used 2, so it in theory should have higher frequency extension than 3 or more output mosfets in parallel.
The Krell uses 1.9 Amps of bias, while off the top of my head, I do not recall what mine is, but I am sure it was more like 2.3 or 2.5 amps of bias, so that could explain the additional spaciousness. My power line I measured with an AC RMS meter and it was drawing 4.7 Amps continuously at roughly 117 volts mains for my Aleph 3. Someone want to do the math to how much bias am I pushing? Is that 275 watts per channel minus transformer losses etc? I would guess 250 watts per channel for an Aleph 3 is quite a bit.
Hopefully, it will last with the fan cooling at 55 to 60 C.
Usually more bias keeps the mosfets more linear, which sounds better. So per ratio, his 50 Watt amp is using less bias per ratio compared to my 30 Watt Aleph.
The aleph would be a second harmonic amp, while the Krells push pull would be 3rd order harmonics? Two styles of sound.
I have tried my Aleph bridged to mono, which cancels the second harmonics, which I have grown fond of. It also rolled the highs off noticeably with back then essentially 4 mosfets paralleled.
Btw, there really was not that many people at this shoot out, so that is why you are not hearing very many opinions. My brother has not responded, because he is on a month long business trip.
I don't know about Lyndon though. He would be the only other person there to comment.
Regards, Bill
Bob
I am not too worried about this test Aleph protype. I have at least 5 more sets of Aleph boards to use for future projects.
I eventually plan on multiamping to go active on my crossover.
I will have to say I have caught the fever for building amps. It has become a lot of fun, since I really enjoy listening to my projects afterwards. I do really appreciate this forum and all of the useful information I have found on this site.
Regards, Bill
I am not too worried about this test Aleph protype. I have at least 5 more sets of Aleph boards to use for future projects.
I eventually plan on multiamping to go active on my crossover.
I will have to say I have caught the fever for building amps. It has become a lot of fun, since I really enjoy listening to my projects afterwards. I do really appreciate this forum and all of the useful information I have found on this site.
Regards, Bill
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.