I built the voltage regulator that Dick refers to (Audio Amateur) with some modifications. There are several posts about it within this thread.
There was a voltage surge problem at turn-on, solved with a zener diode on each LM337 or LM317. The VR fits within the Hafler DH220, but it has a toroid transformer.
Some sound testing (one channel regulated, one not) indicated that the VR did not make much difference, but the speakers (JBL L100) may be easy to drive. There is also 10K uf on each rail (40k total). This may have been enough to drive the MOSFETS and the boards without voltage sag.
Nelson Pass has an article on his site for testing/matching MOSFETS. The simpler one he presents is easy to make and good enough for matching.
There was a voltage surge problem at turn-on, solved with a zener diode on each LM337 or LM317. The VR fits within the Hafler DH220, but it has a toroid transformer.
Some sound testing (one channel regulated, one not) indicated that the VR did not make much difference, but the speakers (JBL L100) may be easy to drive. There is also 10K uf on each rail (40k total). This may have been enough to drive the MOSFETS and the boards without voltage sag.
Nelson Pass has an article on his site for testing/matching MOSFETS. The simpler one he presents is easy to make and good enough for matching.
"Tedious and messy but probably necessary when mounting 6 devices where there were 4. While I'm there I might as well mill out some nice blanking panels for the end of the chassis where I'm removing one heat sink."
I was wondering what to do about blanking the ends, sounds like you have it covered (literally). Be nice to do a 3/8" aluminum plate and cnc something like Hafler Monoblock or similar on it with the correct font. If everything is done up in black you can even use some pseudo gold or brass plated fasteners to give it a boutique look too. Might be a little moneymaker for someone with the time and facilities, to help pay for some other project components.
I believe Dick's monoblock photos show that he managed to locate the extra mosfets between the existing ones, he moved the thermal cutoffs to main bodies of the heat sink itself and bolted through in the spacing between the fins on the exterior.
I ordered the 1981 back issues of Audio Amateur to get a good clean copy of the 81 Pooge article, and found that some of the other issues had related articles as well. Articles by Jung and Boak on regulation are covered, is it safe to assume that the information in them is still fairly relevant today?
I managed to pick up another stock DH220 the other day, and now have 2 of them to work with. I understand Dick that you have had some experience with the Musical Concepts boards, and liked their performance. Given a choice now, would you build monoblocks based on them, or go the modded DH220 board route with the extra Mosfets. Even thought the Musical Concept boards are a little pricey, I could probably sell off the stock 220 boards and components to help offset some of the cost.
Mike
I was wondering what to do about blanking the ends, sounds like you have it covered (literally). Be nice to do a 3/8" aluminum plate and cnc something like Hafler Monoblock or similar on it with the correct font. If everything is done up in black you can even use some pseudo gold or brass plated fasteners to give it a boutique look too. Might be a little moneymaker for someone with the time and facilities, to help pay for some other project components.
I believe Dick's monoblock photos show that he managed to locate the extra mosfets between the existing ones, he moved the thermal cutoffs to main bodies of the heat sink itself and bolted through in the spacing between the fins on the exterior.
I ordered the 1981 back issues of Audio Amateur to get a good clean copy of the 81 Pooge article, and found that some of the other issues had related articles as well. Articles by Jung and Boak on regulation are covered, is it safe to assume that the information in them is still fairly relevant today?
I managed to pick up another stock DH220 the other day, and now have 2 of them to work with. I understand Dick that you have had some experience with the Musical Concepts boards, and liked their performance. Given a choice now, would you build monoblocks based on them, or go the modded DH220 board route with the extra Mosfets. Even thought the Musical Concept boards are a little pricey, I could probably sell off the stock 220 boards and components to help offset some of the cost.
Mike
I, too, did a channel comparison, one regulated one not. Did this about 15 years ago. Although I thought there was an improvement in smoothness and small detail in the sound I concluded the slight improvement was not worth the extra fuss and bother. I put in some larger filter caps on the PCB supply rails and let it go at that. As I understand it the PCB circuit is class A with excellent PSRR so even the modest decoupling circuit in the rails (diode and resistor) is sufficient for its power supply.
Speakers are the weakest link in the audio chain so I don't go way overboard on amplifiers. Having said that I did wax ecstatic about the mods made a few months ago to fashion dual mono channels, improve the drive voltage to the outputs and hang a bunch of quality polyprops to bypass everything possible on the PCB (a la POOGE).
My latest project is an open baffle midrange speaker system, like the Aethers documented at the PE tech support board. I am still astonished at the sound stage and detail and sonic excitement they produce.
Pass's matching circuit is for vertical MOSFETs. Will it work with the lateral MOSFETs used in the Hafler amps? What changes in circuit or voltages must be made to make Pass's circuit work with laterals?
Speakers are the weakest link in the audio chain so I don't go way overboard on amplifiers. Having said that I did wax ecstatic about the mods made a few months ago to fashion dual mono channels, improve the drive voltage to the outputs and hang a bunch of quality polyprops to bypass everything possible on the PCB (a la POOGE).
My latest project is an open baffle midrange speaker system, like the Aethers documented at the PE tech support board. I am still astonished at the sound stage and detail and sonic excitement they produce.
Pass's matching circuit is for vertical MOSFETs. Will it work with the lateral MOSFETs used in the Hafler amps? What changes in circuit or voltages must be made to make Pass's circuit work with laterals?
msb64,
I recently did a comparison between my DH-200 monoblocks, the ones with 80K mF capacitance per channel, with an amp that has the MC circuit cards (PA-3B version) and that has the 26k mF caps sold by MC. Both amps do have 6 MOSFETs output per channel.
The sound differences between them were slight but distinguishable. The MC amp will sound better on a Dvorak string quartet whereas the DH-200 monoblocks sound better on Diana Krall's Live in Paris.
The monoblocks have a large sound that seem to have a lot of power available. I know they sure whap the cr$p out of percussive instrument transients. So, it is a trade off of grace vs: grunt, I guess. For the time being I am still going with the DH-200 monoblocks for all listening.
Now, compared to the stock sound of either the DH-200 or DH-220, the MC mods sound very much better. But (I hate to brag) the monoblocks I did are really great. Is this just "builder's bias" operant? One of these days I will split the MC amp onto two separate chassis and make their power supplies more similar to those of the monoblocks, then a more fair comparison can be made.
So, no definite answer. As so often happens we find in sound production a lot is dependent on personal preferences.
I recently did a comparison between my DH-200 monoblocks, the ones with 80K mF capacitance per channel, with an amp that has the MC circuit cards (PA-3B version) and that has the 26k mF caps sold by MC. Both amps do have 6 MOSFETs output per channel.
The sound differences between them were slight but distinguishable. The MC amp will sound better on a Dvorak string quartet whereas the DH-200 monoblocks sound better on Diana Krall's Live in Paris.
The monoblocks have a large sound that seem to have a lot of power available. I know they sure whap the cr$p out of percussive instrument transients. So, it is a trade off of grace vs: grunt, I guess. For the time being I am still going with the DH-200 monoblocks for all listening.
Now, compared to the stock sound of either the DH-200 or DH-220, the MC mods sound very much better. But (I hate to brag) the monoblocks I did are really great. Is this just "builder's bias" operant? One of these days I will split the MC amp onto two separate chassis and make their power supplies more similar to those of the monoblocks, then a more fair comparison can be made.
So, no definite answer. As so often happens we find in sound production a lot is dependent on personal preferences.
I believe Dick used heatsinks from a P230 or XL280 amp because these Haflers used 3 pairs of output mosfets, so his were already drilled for the third pair of TO3's and the relocated thermal switch. The P230/XL280 (P225 also?) amps both have the extra higher voltage winding in the transformer (unused in the P230).
msb64 said:
I got lucky and found in my metals stock a 9.5" X 9.5" slab of 0.50" aluminum tooling plate. Perfect to saw in half to make two hefty blanking plates to take the places of the removed heat sinks. Then I got in a hurry cutting it and snapped the blade on the band saw. So that's on hold until I make another trip to Production Tool for a new blade.
I plan to make 2 'mirror imaged' monoblocks out of these two DH-200 chassis. They will sit side by side on the shelf with the heat sink ends at the back, the blanking plate ends at the front as faceplates and the RCA inputs, (relocated) power switches, speaker binding posts and line cords at the 'outside' (left side of the left amp and right side of the right amp). That lets me use a single fan to cool the hot ends of both amps. I'll relocate the power transformer to the 'blanking plate' end of the chassis as shown in a previous pic of someone's mod (Dick West?).
This leaves me plenty of room for a separate PSU for the driver boards if I decide I need to.
BTW, another amp that uses boosted supply voltage to the drivers is the Son of Ampzilla. I think it's about 12V/side higher than the output stage's rails. The first time I saw one of those I remember thinking 'what the heck is he up to now?'
Speakers probably are the most critical thing. I've been occupied recently with the EconoWaveGuide development project on AudioKarma. I was responsible for creating the Large Advent conversion project that we just wrapped up:
http://www.audiokarma.org/forums/showthread.php?t=150939
It looks like AudioKarma's crashed at the moment but it'll be back.
I'm planning this Hafler conversion to drive my Linn Isobarik DMS speakers. I've had these for several months and I think they're here to stay. I've had scores of pairs of speakers in this room, from Lowthers to LS3/5As to VOTTs. I like these Linns well enough to go to a little trouble to put together some appropriate amplification for them. Up until now I've been using Accuphase P-300s for everything but the Linns' treble just sounds nicer with MOSFETS so that's the way I'll go now.
Regarding the outputs, I DMM-checked all 16 of the stock ones that I have and I think they're all good. Twelve of them came from 'dead' channels so I expected to find at least a few dead outputs. Must be bad devices on the driver boards. Luckily those are all on hand and cheap.
I'll accept Dick West's generous offer to test and match these devices. If I get 2 usable triplets of each polarity I won't need to use the plastic devices that I bought as replacements.
I still plan to convert each amp from pairs to triplets of devices. The only obstacle to drilling the stock heat sink flanges for a third device between the original two is the mounting holes for the thermal switch on one side. One of the mounting holes would sit under the body of the MOSFET and somewhat reduce contact with the heat sink. I don't know if I care enough about that small amount of missing material to drill, tap, fill and mill off the surface.
Hola you all,
To answer a question in post 719, it is generally accepted that the stock driver boards draw ~50mA each .
Since you have 2 DH200 why don't you stack them top to top
one on the other then you can have both transformers one for each driver board and use all 4 heatsinks and use up to 8 output devices per channell with no drilling. If you're dreaming dream big.
Happy New Year, Elwood
To answer a question in post 719, it is generally accepted that the stock driver boards draw ~50mA each .
Since you have 2 DH200 why don't you stack them top to top
one on the other then you can have both transformers one for each driver board and use all 4 heatsinks and use up to 8 output devices per channell with no drilling. If you're dreaming dream big.
Happy New Year, Elwood
eyoung said:
thanks for the answer on the driver board current.
Using stacked heatsinks/channel is interesting but probably overkill for my application.
Issues:
chassis metalwork becomes complicated.
longer leads to some of the outputs' Gates
overall height
You know, now that I think of it it almost makes sense to do stacked heatsinks on monoblocks.
The driver board would be inside the chassis, flipped upside down so the connections to the outputs would be at the upper edge. The second heatsink would be bolted on top of the first. The connections to the outputs would have to go through a hole cut in the chassis. Some fiddling would have to be done to get the lid to fit. In the upper heat sink the sockets and wiring would be exposed, so some sort of simple cover would have to be applied between the flanges. The cases of the transistors would need plastic covers for safety. In practice I think the N devices would go on one heat sink and the P devices on the other.
I stacked up the heat sinks and chassis parts to see if it was practical. Looks OK if a bit awkward due to height. I took some pix but I don't have them uploaded yet.
Other than those easily managed issues, the only problem is how many outputs the driver board would push without running out of current swing. I know it will drive 6 outputs in the DH-500, but would driving 8 outputs be too much without beefing up Q12/13, the drivers? In a BJT output stage the drive current is largely determined by the output current of the output stage, but I don't have a handle on how drive requirements change in a MOSFET output stage.
Interesting point, Dick about testing lateral versus vertical (or HEXFET) MOSFETs.
From what I can quickly gather, there would be no difference in the simple tests that Nelson Pass mentions. In circuits there are bigger considerations such as gate capacitance and temperature coefficients (the verticals draw more current at they get hotter like bjt's), gate threshold voltages, ringing, etc. In other words, driver circuits would need to be different.
I wonder how Hafler tested their MOSFET's. Is the meter on your tester an ammeter (it's too blurry to tell)? My guess is that the gate switch on the tester is a variable resistor that varies the G-S voltage and the ammeter measures the S-D current. It seems that Hafler would have used some reference S-D current and noted the gate resistance (voltage) class (1 thru 7) at the reference current and stamped this on the device. I could figure this out using some stamped devices. (I have 3's and 5's). Is there anythng on your tester that indicates a reference S-D current?
I matched my lateral MOSFETs on Vgs with a constant S-D current. This would be the reverse of the one I think Hafler used assuming they had a fixed S-D current reference, assuming that the relationship between Vgs and Ids is constant no matter which one is manipulated and under similar device temperatures.
From what I can quickly gather, there would be no difference in the simple tests that Nelson Pass mentions. In circuits there are bigger considerations such as gate capacitance and temperature coefficients (the verticals draw more current at they get hotter like bjt's), gate threshold voltages, ringing, etc. In other words, driver circuits would need to be different.
I wonder how Hafler tested their MOSFET's. Is the meter on your tester an ammeter (it's too blurry to tell)? My guess is that the gate switch on the tester is a variable resistor that varies the G-S voltage and the ammeter measures the S-D current. It seems that Hafler would have used some reference S-D current and noted the gate resistance (voltage) class (1 thru 7) at the reference current and stamped this on the device. I could figure this out using some stamped devices. (I have 3's and 5's). Is there anythng on your tester that indicates a reference S-D current?
I matched my lateral MOSFETs on Vgs with a constant S-D current. This would be the reverse of the one I think Hafler used assuming they had a fixed S-D current reference, assuming that the relationship between Vgs and Ids is constant no matter which one is manipulated and under similar device temperatures.
Please excuse a few "butt-in' questions, as collecting bits to upgrade my recent HD-220 acquisition (original build).
Will this unit function properly if when replacing supply caps, add a C-multiplier to the rails, hence (!) reducing voltage to about 60 volts? Any undersirable Bias Problems, etc (only need a few watts with 97dB speakers so loss of power not a problem)
If okay, is new "Star earth" point at C-Multiplier0volt or original position at supply/ripple caps? (CL-60 to chassis isolating)
Is it better to replace the old resistors with Phillips, Beyschlag, etc, metal films?
Are the little clear plastic Styrene caps for 68pf, 150pF, etc okay to use as replacements, or Silver mica's?
Moving the centre positioned input sockets, away from AC noise and shorter leads, to brackets on heatsinks (top cover clearance hole).
Also intend fitting BYW27 diode bridge and R-C-R-C supply, as per Class A designs - comments?
It appears will need seperate case for supply, no problem.
Fab,
Is there a complete pcb with your extras included?
Thank you ...
Will this unit function properly if when replacing supply caps, add a C-multiplier to the rails, hence (!) reducing voltage to about 60 volts? Any undersirable Bias Problems, etc (only need a few watts with 97dB speakers so loss of power not a problem)
If okay, is new "Star earth" point at C-Multiplier0volt or original position at supply/ripple caps? (CL-60 to chassis isolating)
Is it better to replace the old resistors with Phillips, Beyschlag, etc, metal films?
Are the little clear plastic Styrene caps for 68pf, 150pF, etc okay to use as replacements, or Silver mica's?
Moving the centre positioned input sockets, away from AC noise and shorter leads, to brackets on heatsinks (top cover clearance hole).
Also intend fitting BYW27 diode bridge and R-C-R-C supply, as per Class A designs - comments?
It appears will need seperate case for supply, no problem.
Fab,
Is there a complete pcb with your extras included?
Thank you ...
DH-220 upgrade
Hi
Just a thought: If you have 97dB speaker you probably do not need that much power as the original DH-220. Assuming that 100W is enough for 91-92 dB speakers, you may only need about 25W. So you could go full class A with the actual DH-220 heatsink! You would need something like a 18VAC (25Vdc) transfo and with your capacitance multiplier that would be problably a great sounding amp. I know that Nelson Pass suggested to go class A for this amp by reducing the supply voltage. You may need to increase the drive current of the mosfets to match with class A bias operation but that is feasible if you increase the size of the heatsink of the driver bjts. And you would not need a separate case for the supply (check my 25W class A output mosfet-diff input- JLH converted amp into a DH-200 chassis...)
For a complete pcb with my extras, unfortunately I began a prototype with even more extras but never completed it then I decided to use another bigger chassis to go more into class A- class A/B. However, adding the small daughter boards on the existing Hafler pcb is not that difficult and is very cheap.
I forgot to mention that you could still use a higher regulated voltage for the front end to avoid losing maximum swing voltage at reduced power supply voltage .
Good luck
jameshillj said:
Hi
Just a thought: If you have 97dB speaker you probably do not need that much power as the original DH-220. Assuming that 100W is enough for 91-92 dB speakers, you may only need about 25W. So you could go full class A with the actual DH-220 heatsink! You would need something like a 18VAC (25Vdc) transfo and with your capacitance multiplier that would be problably a great sounding amp. I know that Nelson Pass suggested to go class A for this amp by reducing the supply voltage. You may need to increase the drive current of the mosfets to match with class A bias operation but that is feasible if you increase the size of the heatsink of the driver bjts. And you would not need a separate case for the supply (check my 25W class A output mosfet-diff input- JLH converted amp into a DH-200 chassis...)
For a complete pcb with my extras, unfortunately I began a prototype with even more extras but never completed it then I decided to use another bigger chassis to go more into class A- class A/B. However, adding the small daughter boards on the existing Hafler pcb is not that difficult and is very cheap.
I forgot to mention that you could still use a higher regulated voltage for the front end to avoid losing maximum swing voltage at reduced power supply voltage .
Good luck
dh220 mods started
O.K. I have begun ripping into a dh220 that was collecting dust
it now has shoehorned in two 36vct 6.8a toroids stacked,and a pair of 56,000uf 63v caps . fixed all the goofy grounding and installed a copper ground bar. The front panel will be replaced by a vertical finned heatsink with fins milled away to clear the power switch. have not increased driver current yet.
inrush handled by a CL60 thermister.
So here's the problem right channel bias is set to 450ma but the left will only go to about 350ma with the trimmer maxxed out.
Thinking that maybe I should change out the bias reg transistor.
Also does anyone have the schematic for the hafler matcher, I have a supply of the hitachi mosfets but do not have access to the matcher That I used back when I serviced these things in the 80's. I need to find matches for the ones installed in the amp to mount up the third pair.
O.K. I have begun ripping into a dh220 that was collecting dust
it now has shoehorned in two 36vct 6.8a toroids stacked,and a pair of 56,000uf 63v caps . fixed all the goofy grounding and installed a copper ground bar. The front panel will be replaced by a vertical finned heatsink with fins milled away to clear the power switch. have not increased driver current yet.
inrush handled by a CL60 thermister.
So here's the problem right channel bias is set to 450ma but the left will only go to about 350ma with the trimmer maxxed out.
Thinking that maybe I should change out the bias reg transistor.
Also does anyone have the schematic for the hafler matcher, I have a supply of the hitachi mosfets but do not have access to the matcher That I used back when I serviced these things in the 80's. I need to find matches for the ones installed in the amp to mount up the third pair.
Re: dh220 mods started
Hi
36Vac gives about 50Vdc. With the actual DH-220 heatsink the bias of 450ma may be too much to safely touch the heatsink...
But you mentioned an added heatsink in place of the front panel so it could help provided that you have a good contact with the original heatsink...
To get more than 350ma bias, just change the value of the divider resistors of the bias circuit instead...
Good luck
amp-guy said:
Hi
36Vac gives about 50Vdc. With the actual DH-220 heatsink the bias of 450ma may be too much to safely touch the heatsink...
But you mentioned an added heatsink in place of the front panel so it could help provided that you have a good contact with the original heatsink...
To get more than 350ma bias, just change the value of the divider resistors of the bias circuit instead...
Good luck
yeah I could change the values in the bias divider but this is only in one channel, the other one can go as far as 510ma.
So Iam suspecting that the bias transistor is way off , they are just a pn2222 which have huge beta variations unit to unit.Of course we are talking about quiescent levels that are way out of what was originally intended. Actually even without the extra sink it dosent get all that hot , I ran it all day today with no signal and the sinks were not uncomfortably hot I monitored current with two fluke 98 meters and it was stable.
supposedly the "sweet spot for these fets is 100ma so that =400ma plus 50ma for the vas so this is my target value. Though adding the third pair of outputs would take this up to 650ma that may be a problem. dropping the supply to +-50volts will help dissapation a bit. My plans are to scrapp the case top and fill in across the back with another sink with a piece of punched metal for a top.
The compact chassis is a real challenge to modd these things.
with the added sinks it kind of resembles a Aleph with dwarfish tendencies.
this has been kind of fun so far as it has been a no budget project, everthing comes from the junk box.
So Iam suspecting that the bias transistor is way off , they are just a pn2222 which have huge beta variations unit to unit.Of course we are talking about quiescent levels that are way out of what was originally intended. Actually even without the extra sink it dosent get all that hot , I ran it all day today with no signal and the sinks were not uncomfortably hot I monitored current with two fluke 98 meters and it was stable.
supposedly the "sweet spot for these fets is 100ma so that =400ma plus 50ma for the vas so this is my target value. Though adding the third pair of outputs would take this up to 650ma that may be a problem. dropping the supply to +-50volts will help dissapation a bit. My plans are to scrapp the case top and fill in across the back with another sink with a piece of punched metal for a top.
The compact chassis is a real challenge to modd these things.
with the added sinks it kind of resembles a Aleph with dwarfish tendencies.
this has been kind of fun so far as it has been a no budget project, everthing comes from the junk box.