another place to get a visual of what ticknpop describes is to look at the schematic for hafler's xl280: http://www.hafler.com/techsupport/pdf/XL-280_amp_man.pdf
the input jfets (q1-q4) are "cascoded" by small signal bipolars (q5-q8).
of course, more "improvements" to the dh-200 amp would be to make it even more like the xl-280. good luck finding those jfets, though ...
mlloyd1
the input jfets (q1-q4) are "cascoded" by small signal bipolars (q5-q8).
of course, more "improvements" to the dh-200 amp would be to make it even more like the xl-280. good luck finding those jfets, though ...
mlloyd1
another place to get a visual of what ticknpop describes is to look at the schematic for hafler's xl280: http://www.hafler.com/techsupport/pdf/XL-280_amp_man.pdf
the input jfets (q1-q4) are "cascoded" by small signal bipolars (q5-q8).
of course, more "improvements" to the dh-200 amp would be to make it even more like the xl-280. good luck finding those jfets, though ...
mlloyd1
Hi,
thanks for the tip! I see better now what ticknpop meant!
Anyhow, is the DH-280 really an improvement respect to DH-2X0s?
Ciao
Paolo
i had a chance many years ago to listen to them side by side on some nice b&w speakers. the low ends were similar, but the mids and highs sounded more pleasing to me on the xl-280, your mileage may vary.
just for fun, we also did the subtraction test described in the xl-280 manual. the results were strange. spec-wise, the dh-220 has lower thd, but the xl-280 produced the best null.
must be the jfets ...
(a joke)
mlloyd1
just for fun, we also did the subtraction test described in the xl-280 manual. the results were strange. spec-wise, the dh-220 has lower thd, but the xl-280 produced the best null.
must be the jfets ...
(a joke)
mlloyd1
matching 2N5551/2N5401 in VAS on a Hafler XL-280
Hi pidigi,
I 'had' a factory-built XL-280 that I started messing with and before I knew it, I had replaced the 2N5550/2N5401 transistors on one channels VAS, but not the other.
When I did it, I wanted to do it right so I went with ONSemi 2N5551's in place of the original Mitsubishi 2N5550's. I also used ONSemi 2N5401's.
Before installation, I matched them using a cheap DMM with a transistor hFe function. I bought a lot of 20 apiece and they were all pretty close. I found 2 2N5551 and 2 2N5401 that measured the same at room temp and warmed to body temp using my fingers to loosely pinch them until the reading stabilized. As a side note, the original transistors measured all over the place.
The Result:
Not much of a difference.
The offset was -13.9mV before the operation, -14.1 after.
The 'new' side is a bit brighter, maybe a bit more dynamic sounding, but the old side is very balanced sounding with a bit more bass. Comes down to preference in my case.
What blows is that I will have to do the other side one of these days, the original mission was and always has been to create the most 'perfect' XL-280 possible because, for me, it is a classic circuit that has given me hours of musical amazement, I believe in this circuit, if you will.
Never touch a good-sounding original amp, they're like cars with matching numbers....
but it was fun
Hi pidigi,
I 'had' a factory-built XL-280 that I started messing with and before I knew it, I had replaced the 2N5550/2N5401 transistors on one channels VAS, but not the other.
When I did it, I wanted to do it right so I went with ONSemi 2N5551's in place of the original Mitsubishi 2N5550's. I also used ONSemi 2N5401's.
Before installation, I matched them using a cheap DMM with a transistor hFe function. I bought a lot of 20 apiece and they were all pretty close. I found 2 2N5551 and 2 2N5401 that measured the same at room temp and warmed to body temp using my fingers to loosely pinch them until the reading stabilized. As a side note, the original transistors measured all over the place.
The Result:
Not much of a difference.
The offset was -13.9mV before the operation, -14.1 after.
The 'new' side is a bit brighter, maybe a bit more dynamic sounding, but the old side is very balanced sounding with a bit more bass. Comes down to preference in my case.
What blows is that I will have to do the other side one of these days, the original mission was and always has been to create the most 'perfect' XL-280 possible because, for me, it is a classic circuit that has given me hours of musical amazement, I believe in this circuit, if you will.
Never touch a good-sounding original amp, they're like cars with matching numbers....
but it was fun
Hello HaflerFreak,
Thanks for the answer. Thats more or less my philosophy, I like the DH-200 and I'm trying to squeeze out from them as much as possible. I've built some "clones" and also refurbished some units. Yes, never touch a good sounding amp, but after more than 20 years something has to be done
On the clones and the stock amps I use to substitute all the capacitors and also all the resistors, using wima and other mkp and metal film resistors. I throw away all the old cabling, put a mains socket, rewire the ground path. In the clones I have noticed the importance of the input bjt matching (also soundwise), but I've never tried to experiment with the other bjts.
Ciao,
Paolo
(very nice nikname indeed!)
Thanks for the answer. Thats more or less my philosophy, I like the DH-200 and I'm trying to squeeze out from them as much as possible. I've built some "clones" and also refurbished some units. Yes, never touch a good sounding amp, but after more than 20 years something has to be done
On the clones and the stock amps I use to substitute all the capacitors and also all the resistors, using wima and other mkp and metal film resistors. I throw away all the old cabling, put a mains socket, rewire the ground path. In the clones I have noticed the importance of the input bjt matching (also soundwise), but I've never tried to experiment with the other bjts.
Ciao,
Paolo
(very nice nikname indeed!)
hi haflerfreak:
if you're saying you replaced Q9 and Q10 (VAS emitter followers) with a pair having closer hFE match, the lack of offset change is not so surprising. i suppose it is possible that you might get a little bit of improvement if you matched for vBE at the operating point instead.
although it sure is nice to see "0.0mV", as you know, 14mV is not at all a bad output offset. the 1000uF in the feedback leg is helping a lot as well.
also, as i recall, the xl-280 default is to use 5% (!?!?) tolerance carbon film resistors in most areas, including some key areas that set operating points. we can only imagine what would happen is some of those are off by close to 5% in opposite directions ...
i think you'd get bigger "bang for the buck" by replacing those "sloppy" carbon film resistors with 1% metal film first, then match the semiconductors afterwards. and, i think the jfets and actual VAS transistors and drivers would be more critical to match that the VAS emitter followers and cascodes for the jfets.
just my thoughts; ymmv ...
good luck and take care of that xl-280! i agree about not touching a good sounding amp.
mlloyd1
if you're saying you replaced Q9 and Q10 (VAS emitter followers) with a pair having closer hFE match, the lack of offset change is not so surprising. i suppose it is possible that you might get a little bit of improvement if you matched for vBE at the operating point instead.
although it sure is nice to see "0.0mV", as you know, 14mV is not at all a bad output offset. the 1000uF in the feedback leg is helping a lot as well.
also, as i recall, the xl-280 default is to use 5% (!?!?) tolerance carbon film resistors in most areas, including some key areas that set operating points. we can only imagine what would happen is some of those are off by close to 5% in opposite directions ...
i think you'd get bigger "bang for the buck" by replacing those "sloppy" carbon film resistors with 1% metal film first, then match the semiconductors afterwards. and, i think the jfets and actual VAS transistors and drivers would be more critical to match that the VAS emitter followers and cascodes for the jfets.
just my thoughts; ymmv ...
good luck and take care of that xl-280! i agree about not touching a good sounding amp.
mlloyd1
Hi pidigi,
I 'had' a factory-built XL-280 that I started messing with and before I knew it, I had replaced the 2N5550/2N5401 transistors on one channels VAS, but not the other.
...
Never touch a good-sounding original amp, they're like cars with matching numbers....
but it was fun
I'll chime in and agree that replacing all of the carbon films with closely matched metal films is a very worthwhile upgrade. The signal path resistors and all of the resistors associated with the input stage and feedback circuit should get the best quality parts you can afford; 100 ppm Tc resistors are okay, but 50 ppm or even 25 ppm should be better.
I'm in the process of repairing and upgrading a friend's Son of Ampzilla, including using better quality electrolytics, especially a proper bipolar in the feedback circuit. Always used Black Gate NXs here, but they're becoming unobtanium and I couldn't afford to stock up on a lifetime supply.
Who wants some 2k ohm Vishay bulk metal foil resistors, probably .01% tolerance and at least 5 ppm Tc? I have 99, about half of which I'm probably going to give away. Got the whole batch for $14 off Ebay. PM with a mailing addy.
I'm in the process of repairing and upgrading a friend's Son of Ampzilla, including using better quality electrolytics, especially a proper bipolar in the feedback circuit. Always used Black Gate NXs here, but they're becoming unobtanium and I couldn't afford to stock up on a lifetime supply.
Who wants some 2k ohm Vishay bulk metal foil resistors, probably .01% tolerance and at least 5 ppm Tc? I have 99, about half of which I'm probably going to give away. Got the whole batch for $14 off Ebay. PM with a mailing addy.
Hi Mlloyd,
Of course you are right, it has been awhile since my XL-280 project.
I replaced Q5, Q6, Q7, Q8, and Q9 and Q10, so, the cascodes also. All the transistors were ONSemi from the same batch, I believe the 5401's were a point or two below the 5551's but all six were virtually identical beta on my DMM. The new transistors had a lower beta than the old ones, but as I mentioned, the old ones were obviously not matched.
An interesting mod note: The 6.3V 1000uF BP electro in the feedback loop was originally a Nichicon BP in a big can. I tried many replacements, none have sounded as good as the original caps which I threw away but the best I have found is, ironically, the only Nichicon BP in the identical can size, the beautiful 16V 1000uF BP from Nichicon (I warn you about trying Panasonic BPs here, the worst!). Works well.
I haven't touched the 2N3440 or 2N5415, but I have a couple laying around if there's trouble.
Resistors:
I have replaced the critical resistors on both driver cards with closely matched Dale RN60D resistors. That would be R1, R2, R25, R26, R27.
Capacitors:
Replaced a lot of caps with Panasonic polyprops, electros for driver card PS increased from 470uF to 1000uF Nichicon FW's. New diodes in the bridge for the drivers.
I guess I could go on and on, I need to dig up my pics - some of which may have been lost.
Haven't heard her in a looong time, but that's kind of why I'm back on the board, I am no longer semi-homeless, living in a dining room -- got on a lease and I am looking forward to long listening sessions and winter projects
Unfortunately, I recall the sound becoming a bit flat when the amp got hot after I did all the mods, but I had a lousy listening room at my last apartment. My new room should be better acoustically.
Gotta figure out how to move the Maggies from storage, but I'll get there.
I'll try to post some pics if I can find them, or new ones when I get the Haffie moved in....
Of course you are right, it has been awhile since my XL-280 project.
I replaced Q5, Q6, Q7, Q8, and Q9 and Q10, so, the cascodes also. All the transistors were ONSemi from the same batch, I believe the 5401's were a point or two below the 5551's but all six were virtually identical beta on my DMM. The new transistors had a lower beta than the old ones, but as I mentioned, the old ones were obviously not matched.
An interesting mod note: The 6.3V 1000uF BP electro in the feedback loop was originally a Nichicon BP in a big can. I tried many replacements, none have sounded as good as the original caps which I threw away
but the best I have found is, ironically, the only Nichicon BP in the identical can size, the beautiful 16V 1000uF BP from Nichicon (I warn you about trying Panasonic BPs here, the worst!). Works well.I haven't touched the 2N3440 or 2N5415, but I have a couple laying around if there's trouble.
Resistors:
I have replaced the critical resistors on both driver cards with closely matched Dale RN60D resistors. That would be R1, R2, R25, R26, R27.
Capacitors:
Replaced a lot of caps with Panasonic polyprops, electros for driver card PS increased from 470uF to 1000uF Nichicon FW's. New diodes in the bridge for the drivers.
I guess I could go on and on, I need to dig up my pics - some of which may have been lost.
Haven't heard her in a looong time, but that's kind of why I'm back on the board, I am no longer semi-homeless, living in a dining room -- got on a lease and I am looking forward to long listening sessions and winter projects
Unfortunately, I recall the sound becoming a bit flat when the amp got hot after I did all the mods, but I had a lousy listening room at my last apartment. My new room should be better acoustically.
Gotta figure out how to move the Maggies from storage, but I'll get there.
I'll try to post some pics if I can find them, or new ones when I get the Haffie moved in....
Hi HaflerFreak
the DH-200 doesn't come with the Nichicon... I have replaced the original BP with a NITAI, never ever thougth about trying other fancy brands, but now your informations make me very curious and I'd like to give the Nichicon a try! When you speak about the Nichicon BP, are you referring to the MUSE ES green model? The other question is about voltage. It seems you are suggesting the 16V instead of the original 6.3V: is it really better?
...interesting! I suppose you have choosen a soft recovery diode, right?
yes please
I find it very useful to exchange informations about Hafler tweaking!
wow, Maggies and Hafler sounds like a neat combination!
Ciao
Paolo
.....
An interesting mod note: The 6.3V 1000uF BP electro in the feedback loop was originally a Nichicon BP in a big can. I tried many replacements, none have sounded as good as the original caps which I threw away
the DH-200 doesn't come with the Nichicon... I have replaced the original BP with a NITAI, never ever thougth about trying other fancy brands, but now your informations make me very curious and I'd like to give the Nichicon a try! When you speak about the Nichicon BP, are you referring to the MUSE ES green model? The other question is about voltage. It seems you are suggesting the 16V instead of the original 6.3V: is it really better?
...interesting! I suppose you have choosen a soft recovery diode, right?
yes please
I find it very useful to exchange informations about Hafler tweaking!
wow, Maggies and Hafler sounds like a neat combination!
Ciao
Paolo
did anyone try this?
I have seen this tweak on another thread:
http://www.diyaudio.com/forums/solid-state/152916-hafler-dh-200-bias-problem.html
and I'm wondering if someone else tried this tweak.
Ciao
Paolo
I have seen this tweak on another thread:
http://www.diyaudio.com/forums/solid-state/152916-hafler-dh-200-bias-problem.html
and I'm wondering if someone else tried this tweak.
Ciao
Paolo
Hi folks, one more improvement on the DH-200 worth mentioning. I took my high dollar Denon to compare to the DH-200 and found it had an advantage in the mids. So I took the DH-200 back to the bench and added a 100uF 35V Nichicon Muse FW/KW/KZ (I forget the series) eletrolytic cap in parallel with the 10 uF bipolar Nichicon MUSE ES cap (positieve side to the input side). It brought the mids right out in a great way. Took it back and A/B with the Denon. The Hafler DH-200 kicked its butt quite easily. The DH-200 sounds natural, the Denon sonds a bit megaphonish through my horns. Before this bypass, the mids where a bit costrained, now they have opened up and sound better to my ears than the Denon, and the Denon is respectable circuity, at least in the power amp. The preamp of the Denon is just OK.
If any of you try it, let me know the results.
If any of you try it, let me know the results.
Hi
I have a DH-220 that I need to add a input level control for one channel. I have two sub-woofers one being closer to a corner and need to balance it out with the other one. I have a LFE mono output from my preamp that I am using a "Y" adapter to split into the two channels. What will work best and do you have a schematic for it. Thanks
I have a DH-220 that I need to add a input level control for one channel. I have two sub-woofers one being closer to a corner and need to balance it out with the other one. I have a LFE mono output from my preamp that I am using a "Y" adapter to split into the two channels. What will work best and do you have a schematic for it. Thanks
Hi,
yes, it seems that the Hafler guys are silent lately
About your question, it will do no harm if you put this potentiometer, but I'm not sure this will solve your problem.
If the crossover frequency is reasonably low, you shouldn't be able to localize the position of the subwoofer, so it is not clear what you mean when you say that you need to balance them. Or do you have nasty interference between them?
Ciao
Paolo
P.S.: I'm building a new Hafler with two boards that I have made the layout. Gold plated traces, six mosfets per channel (2SK1058-2SJ162). I'm going to regulate the voltage for the "signal" side (I have two additionals windings on the transformer), and I would like to use fast (or soft) recovery diodes in the power supply. I'll post more pictures and results as soon as they are available.
yes, it seems that the Hafler guys are silent lately
About your question, it will do no harm if you put this potentiometer, but I'm not sure this will solve your problem.
If the crossover frequency is reasonably low, you shouldn't be able to localize the position of the subwoofer, so it is not clear what you mean when you say that you need to balance them. Or do you have nasty interference between them?
Ciao
Paolo
P.S.: I'm building a new Hafler with two boards that I have made the layout. Gold plated traces, six mosfets per channel (2SK1058-2SJ162). I'm going to regulate the voltage for the "signal" side (I have two additionals windings on the transformer), and I would like to use fast (or soft) recovery diodes in the power supply. I'll post more pictures and results as soon as they are available.
Attachments
Paolo
Thanks for the reply.
The two transmission line subwoofers are mounted between the floor joists in my great room and cannot be relocated. Since one subwoofer driver and port is located closer to a corner it amplifies the bass and the level needs to be reduced to match the output sound level of the other subwoofer. My new processor only has a single LFE output so I need to be able to reduce the output level of the subwoofer in the corner so both subwoofers are at the same volume level when measured with a SPL sound meter located where I am sitting. Hope that helps.
Thanks for the reply.
The two transmission line subwoofers are mounted between the floor joists in my great room and cannot be relocated. Since one subwoofer driver and port is located closer to a corner it amplifies the bass and the level needs to be reduced to match the output sound level of the other subwoofer. My new processor only has a single LFE output so I need to be able to reduce the output level of the subwoofer in the corner so both subwoofers are at the same volume level when measured with a SPL sound meter located where I am sitting. Hope that helps.
C9 and parallel resistor
Hi Paolo
C9 is not a Cdom but a "shunt lag" compensation (VAS shunt to ground). The DH-200 does not use Cdom Miller cap but C9 and RC networks in the collector load of the input stage for frequency compensation scheme. Some schools of thoughts say that Cdom is not good for audio and some other schools say that shunt compensation (C and/or R to shunt VAS to GND) loads to much the VAS stage. The DH-200 frequency compensation scheme uses a small C9 and the rest of the compensation (except feedback compensation) is done in input stage. So, it may be one reason why the DH-200 sounds good. Putting a 20k in parralel extend the open loop bandwidth of the amp but at the price of Power supply ripple increase. So, there is no free lunch.
However, 20K is a high value which should limit the perpectible effect on ripple generation.
Fab
Hi Paolo
C9 is not a Cdom but a "shunt lag" compensation (VAS shunt to ground). The DH-200 does not use Cdom Miller cap but C9 and RC networks in the collector load of the input stage for frequency compensation scheme. Some schools of thoughts say that Cdom is not good for audio and some other schools say that shunt compensation (C and/or R to shunt VAS to GND) loads to much the VAS stage. The DH-200 frequency compensation scheme uses a small C9 and the rest of the compensation (except feedback compensation) is done in input stage. So, it may be one reason why the DH-200 sounds good. Putting a 20k in parralel extend the open loop bandwidth of the amp but at the price of Power supply ripple increase. So, there is no free lunch.
However, 20K is a high value which should limit the perpectible effect on ripple generation.
Fab