Is it worth the effort to locate non-ferrous mounting hardware for toroidal transformers for this application?
It is very easy to fabricate the necessary large retaining washers with plate aluminum and a 3.75" hole saw bit.
Most hardware stores should have solid brass toilet tank bolts that are 5/16" x 3" long, and some even come with brass washers and nuts as well as rubber washers too.
It is very easy to fabricate the necessary large retaining washers with plate aluminum and a 3.75" hole saw bit.
Most hardware stores should have solid brass toilet tank bolts that are 5/16" x 3" long, and some even come with brass washers and nuts as well as rubber washers too.
Hi amp-guy,
The way I do this is the measure a bunch of mosfets and group them by gate voltage for a specified current with about 10 VDC on the drain. Once I have this done, I mount like parts on a jig I made up that keep each transistor closer in temperature to the others. With a common gate voltage, I measure each value of drain current. You have to do this once the assembly reaches thermal equilibrium. This set up is capable of matching mosfets close enough for something like a Counterpoint (no source resistors to force current sharing). You can also look at the amount of spread that your earlier matches have so you have an idea how close you were (or were not possibly!). This would give you an acceptable window, or acceptable spread in Vgs.
Watch your transformer core temperature! It can take hours to reach the temperature they will run at. Run the amp at some power level and watch the temperature. Inch your way up in power output, or bias current, until you are at the level you can accept. The core temperature must stay below the temperature of the thermal fuse buried in the windings. It opens permanently at a temperature. Bye, bye transformer. It is in your own best interest to make sure you have headroom between your worst case operating temperatures and the temperature of parts you will generate. The power transformer is only one part. There are others as well.
-Chris
Look at Nelson Pass' web site. Matching mosfets is pretty straight forward, but you do have to compensate for temperature.
The way I do this is the measure a bunch of mosfets and group them by gate voltage for a specified current with about 10 VDC on the drain. Once I have this done, I mount like parts on a jig I made up that keep each transistor closer in temperature to the others. With a common gate voltage, I measure each value of drain current. You have to do this once the assembly reaches thermal equilibrium. This set up is capable of matching mosfets close enough for something like a Counterpoint (no source resistors to force current sharing). You can also look at the amount of spread that your earlier matches have so you have an idea how close you were (or were not possibly!). This would give you an acceptable window, or acceptable spread in Vgs.
Watch your transformer core temperature! It can take hours to reach the temperature they will run at. Run the amp at some power level and watch the temperature. Inch your way up in power output, or bias current, until you are at the level you can accept. The core temperature must stay below the temperature of the thermal fuse buried in the windings. It opens permanently at a temperature. Bye, bye transformer. It is in your own best interest to make sure you have headroom between your worst case operating temperatures and the temperature of parts you will generate. The power transformer is only one part. There are others as well.
-Chris
DH-220 heatsink temperature
I measured that one stock DH-200/220 heatsink can dissipate about 55W for about 48 degC at the middle and on top of heatsink. So I guess you could go for about 550ma or so at +/-50Vdc for the mosfet - excluding the driver stage current.
Good luck and keep us posted on your mods. Some pictures would be nice.
amp-guy said:
I measured that one stock DH-200/220 heatsink can dissipate about 55W for about 48 degC at the middle and on top of heatsink. So I guess you could go for about 550ma or so at +/-50Vdc for the mosfet - excluding the driver stage current.
Good luck and keep us posted on your mods. Some pictures would be nice.
well some progress today the heatsink\front panel is machined ready to be deburred and finished.
I took the opportunity to beef up the power switch and changed the location to the middle of the face, this necessitated milling out some fins and making a small enclosure to contain it. The panel\sink will be bonded to the existing sinks through a section of 1/4" wall alu angle with a film of arctic silver applied. milling across fins is tediously slow as very small cuts must be made otherwise the cutter will grab the fin and bend the *^$#^&*out of it.
I took the opportunity to beef up the power switch and changed the location to the middle of the face, this necessitated milling out some fins and making a small enclosure to contain it. The panel\sink will be bonded to the existing sinks through a section of 1/4" wall alu angle with a film of arctic silver applied. milling across fins is tediously slow as very small cuts must be made otherwise the cutter will grab the fin and bend the *^$#^&*out of it.
Attachments
amp-guy said:
For the drivers I would not expect a big improvement if you keep the relatively small actual drive current of the mosfets on the stock DH-200/220.
For the VAS, changing the 2n5415/2n3440 for lower Cob and higher Ft BJT (not the mje15030\mje15031 ) would be more beneficial...I think.
Power required to drive mosfet I/P capacitance
For all those gentlemen that require the facts, straight from Hitachi here is the
derivation of I/P power to drive the mosfet..
Pin=Freq.Ciss.vgs^2...
at 100khz,900pf,6Volts=100x10^3x900x10^-12x6^2=3.24milliwatts..ergo the higher the frequency the higher the drive requirement, so at 10khz the drive will be
0.324milliwatts.. allow for the difference in I/p capacitance between the P, and N
type Mosfets in your calculations..
The types referred to are the 2SK134,2SJ49 and equivelent 2SK1058,and2SJ162.
The 2SJ49 has the CISS I/P capacitance of 900pf abd the 2SK134 has but 600pf
I/P capacitance ..The Hafler 200 drivers run at about 8 to 9m/a, all that is needed is
a power calculation to see how many mosfets this circuit will satisfy, and if more drive is required just lower the 220ohm resistor, but watch the power dissipation
at about 12m/a this will be roughly the same as the predrivers with a similar heat
dissipation. Remember that each set of mosfets will be driven at a time not all at once
as the drivers swing + then -- hope thats all clear
Regards
For all those gentlemen that require the facts, straight from Hitachi here is the
derivation of I/P power to drive the mosfet..
Pin=Freq.Ciss.vgs^2...
at 100khz,900pf,6Volts=100x10^3x900x10^-12x6^2=3.24milliwatts..ergo the higher the frequency the higher the drive requirement, so at 10khz the drive will be
0.324milliwatts.. allow for the difference in I/p capacitance between the P, and N
type Mosfets in your calculations..
The types referred to are the 2SK134,2SJ49 and equivelent 2SK1058,and2SJ162.
The 2SJ49 has the CISS I/P capacitance of 900pf abd the 2SK134 has but 600pf
I/P capacitance ..The Hafler 200 drivers run at about 8 to 9m/a, all that is needed is
a power calculation to see how many mosfets this circuit will satisfy, and if more drive is required just lower the 220ohm resistor, but watch the power dissipation
at about 12m/a this will be roughly the same as the predrivers with a similar heat
dissipation. Remember that each set of mosfets will be driven at a time not all at once
as the drivers swing + then -- hope thats all clear
Regards
A look at the manual and yes the drivers and predivers are the 2n5415\3440 now where did I put the datasheets on all these devices, oh there it is, the 25415 can dissipate a watt and has an Ft of 15mhz that's really not too shabby . But still I wonder if changing to a device characterised for high quality audio might not give better results. Thanks humble 12 ma sounds about what I would have guessed. Like you say as freq goes up gate capacitance becomes more of an issue to drive. I wonder if this all has something to do with the hardness I hear in the highs with hafler amps.fab said:
....please Check again, according to the Hafler Manuel VAS is still a darlington as for the DH-200 and the main current is provided by Q8 and Q11 which are 2n5415/2n3440.
amp-guy said:
The best way is to try it out. On my side, I use double the drive current for mosfet simply adding a 220 ohms in parallel with the existing one - but I install it at the connection point on the hetasink inside the amp at the junction of the mosfet gate resistors connection to avoid soldering/desoldering on the pcb (at least for trial). I also install the 1uF switching speed-up cap there since it is also very convenient.
Better Drivers Predrivers
Hi Ampguy, this question has already been answered in thread [465], suggests
2SC2911 and 2SA1209, 150mhz 160volts, classified as complimentary"audio devices."
There are also comments about the 2N3440/5415 regarding their shortcomings.
Suggest you read the previous writings and the many mods contained therein from
link 16 onwards I know its a drag but many questions are arising that have already been answered..Do get rid of those AWFUL video 3440/5415 devices if you feel
technically able, 15mhz FT is not that good for driving mosfets, and some of those devices had very low gains and matching very difficult between P and N types
maybe reasonable in their time but very dated now.. Changing to the new transistors will also allow a smaller "Dominent pole capacitor" C9, 560pf down to 330pf.. this will speed up the amp and smooth out that treble.The devices mentioned are TO126
and with a suitable heatsink will dissipate several watts this will allow you to increase the gain of the basic driver circuit by changing R27 and R33 to a lower value from 100
ohms down to 82ohms or even lower depending on heatsink size.. all these mods already alluded to in previous links.. regards
Humble
Hi Ampguy, this question has already been answered in thread [465], suggests
2SC2911 and 2SA1209, 150mhz 160volts, classified as complimentary"audio devices."
There are also comments about the 2N3440/5415 regarding their shortcomings.
Suggest you read the previous writings and the many mods contained therein from
link 16 onwards I know its a drag but many questions are arising that have already been answered..Do get rid of those AWFUL video 3440/5415 devices if you feel
technically able, 15mhz FT is not that good for driving mosfets, and some of those devices had very low gains and matching very difficult between P and N types
maybe reasonable in their time but very dated now.. Changing to the new transistors will also allow a smaller "Dominent pole capacitor" C9, 560pf down to 330pf.. this will speed up the amp and smooth out that treble.The devices mentioned are TO126
and with a suitable heatsink will dissipate several watts this will allow you to increase the gain of the basic driver circuit by changing R27 and R33 to a lower value from 100
ohms down to 82ohms or even lower depending on heatsink size.. all these mods already alluded to in previous links.. regards
Humble
Thanks I will have to go back and have a look I lost touch with this thread for quite awhile and it looks like I missed out . I used the 2n5415 about 20 years ago in a .d.c motor constant torque drive circuit you have jogged my memory they did have pretty wide spread on gain. those are pretty significant changes in caps that should make a difference. what about the 2n5401\5551 in the diff input stages Any reccomendations there?
My project here was intended to be a quick and easy diversion using just parts on hand but has gotten out of control what with new transformers and added heatsinking and now transistor swaps . All I was after was a temporary amp to drive my maggies while I cannabolize the current amp to build a new one.
Oh well it's all in fun else why do it.
My project here was intended to be a quick and easy diversion using just parts on hand but has gotten out of control what with new transformers and added heatsinking and now transistor swaps . All I was after was a temporary amp to drive my maggies while I cannabolize the current amp to build a new one.
Oh well it's all in fun else why do it.
Re 2N5401/5551
Hi Ampguy, although these devices are quite old they are so very useable at 160V
0.6A and 100MHZ, fairly rugged and readily available, and like the song made for one another, overall difficult to fault in the configuration as used by D Hafler. They are also favoured by the author of "HIGH POWER AUDIO AMPLIFIER CONSTRUCTION MANUAL" by G RANDY SLOANE of SEAL ELECTRONICS, in which he uses them
prolifically in all his amp designs, both MOS and BI-POLAR.. In short they have withstood the test of time.There are other more modern devices available with slightly lower noise figures like 2SA970/2SC2240 which I have used in I/P
circuits but at 0.1amp much less rugged and you cannot really tell any difference with the overall noise figures of an amplifier after feedback etc, maybe better in discreet preamps where higher gain may be required say from a MM stage..
Best stick with what works and is easy to get cheaply. In any case once you have caught up with all the earlier posts you will begin to see that the I/P devices are really blameless compared to all the other possible improvements that one can do.
Referring back to earlier post about the SANYO AUDIO DRIVERS 2SA1209/2SC2911
They feature the following as AUDIO DRIVERS. LOW DISTORTION,EXCELLENT Hfe
LINEARITY,LOW OUTPUT CAPACITANCE, HIGH BREAKDOWN VOLTAGE and are
recommended for driving mosfets.. Info courtesy www.profusionplc.com
This is the company that supplied me and my colleage when we were doing
a monoblock update version of the D Hafler DH200 using 4 sets of O/P mosfets
producing 350watts, and sounding like SILK...
Regards
Humble
Hi Ampguy, although these devices are quite old they are so very useable at 160V
0.6A and 100MHZ, fairly rugged and readily available, and like the song made for one another, overall difficult to fault in the configuration as used by D Hafler. They are also favoured by the author of "HIGH POWER AUDIO AMPLIFIER CONSTRUCTION MANUAL" by G RANDY SLOANE of SEAL ELECTRONICS, in which he uses them
prolifically in all his amp designs, both MOS and BI-POLAR.. In short they have withstood the test of time.There are other more modern devices available with slightly lower noise figures like 2SA970/2SC2240 which I have used in I/P
circuits but at 0.1amp much less rugged and you cannot really tell any difference with the overall noise figures of an amplifier after feedback etc, maybe better in discreet preamps where higher gain may be required say from a MM stage..
Best stick with what works and is easy to get cheaply. In any case once you have caught up with all the earlier posts you will begin to see that the I/P devices are really blameless compared to all the other possible improvements that one can do.
Referring back to earlier post about the SANYO AUDIO DRIVERS 2SA1209/2SC2911
They feature the following as AUDIO DRIVERS. LOW DISTORTION,EXCELLENT Hfe
LINEARITY,LOW OUTPUT CAPACITANCE, HIGH BREAKDOWN VOLTAGE and are
recommended for driving mosfets.. Info courtesy www.profusionplc.com
This is the company that supplied me and my colleage when we were doing
a monoblock update version of the D Hafler DH200 using 4 sets of O/P mosfets
producing 350watts, and sounding like SILK...
Regards
Humble
You don't need to sell me on the 2n5401\2n5551 I have used lots of them in fact ,as I write Iam listening to a pass A40 with those devices used throughout. A quick check shows they are still active production at Onsemi, Fairchild and Semtech. The 2sa970/2sc2240 are no spring chickens either, lots of Japanese amps have used them since back in the 80's when I sat at a bench repairing amps daily. You really can't go wrong with either one. Iam however not familiar with the 2sa1209\2sc2911 combo
do you think them to be comparable or better than the mje15030\mje15031 pair i actually have about 50 each on hand of NOS motorola of both the mje's and the 2n's but am not adverse to Japanese transistors by any means.
do you think them to be comparable or better than the mje15030\mje15031 pair i actually have about 50 each on hand of NOS motorola of both the mje's and the 2n's but am not adverse to Japanese transistors by any means.
Motorola's 15030/15031
Hi Amp-Guy, Have not used these in a Hafler amp but have been pleased with the
Sanyo's, lets put things into perspective the Sanyo's are fast so when considering
the power of the monoblocks [350watts] I needed a faster slew-rate to cope with the greater voltage swings in my monoblocks..If you are just going to find a replacement
for the 3440/5416 in a standard Hafler Circuit you will be asking about advantages
in changing what is already there, you imply that you would like a smoother treble,
and are basically asking will your 15030/15031 help the situation, the answer is I don't honestly know but a comparison with the basic parameters reveal that the
motorola's are only twice as fast as the 3440/5416 at 30mhz and also have a rather low gain @40 min,hfe, in short not much better than the devices you want to change.
Personally if its just for standard Hafler, I would go ahead and try them, but incorporate a heatsink that will enable you to increase the current/power dissipation
in the predrivers and push the gain up a little I.E R27 and R33 from 100ohms to 82
ohms this will give you more feedback and should help the treble to be a little smoother, for the predrivers select the highest gain matches from your selection, to help reduce pulling from I/P diffs, and thus keeping the offset as low as possible..
If you do take the boards off to attempt this, suggest that at same time you increase the capacitor across amplified diode [C8] to 1 to 2 microfarads 63v poly layers are excellent [original caps are 0.1 @25V] but crappy ceramics which don't sound good.
Over to you sir
regards
Hi Amp-Guy, Have not used these in a Hafler amp but have been pleased with the
Sanyo's, lets put things into perspective the Sanyo's are fast so when considering
the power of the monoblocks [350watts] I needed a faster slew-rate to cope with the greater voltage swings in my monoblocks..If you are just going to find a replacement
for the 3440/5416 in a standard Hafler Circuit you will be asking about advantages
in changing what is already there, you imply that you would like a smoother treble,
and are basically asking will your 15030/15031 help the situation, the answer is I don't honestly know but a comparison with the basic parameters reveal that the
motorola's are only twice as fast as the 3440/5416 at 30mhz and also have a rather low gain @40 min,hfe, in short not much better than the devices you want to change.
Personally if its just for standard Hafler, I would go ahead and try them, but incorporate a heatsink that will enable you to increase the current/power dissipation
in the predrivers and push the gain up a little I.E R27 and R33 from 100ohms to 82
ohms this will give you more feedback and should help the treble to be a little smoother, for the predrivers select the highest gain matches from your selection, to help reduce pulling from I/P diffs, and thus keeping the offset as low as possible..
If you do take the boards off to attempt this, suggest that at same time you increase the capacitor across amplified diode [C8] to 1 to 2 microfarads 63v poly layers are excellent [original caps are 0.1 @25V] but crappy ceramics which don't sound good.
Over to you sir
regards
You make a good point , if you are going to the fuss of removing the board then might as well put in devices that have been tried and proven to be top performers. so I think I will put it off a bit and order up some of the sanyo's. mcminone.com should have them. As for ceramic caps (phooey I get a bad taste just typing the word) this is a DH220 here- all polypropylene oh but wait you are saying increase the value. hmm sounds good I have some 1uf 63 v BC poly's on hand . the rails here are about 53 volts as I have dropped in two 36v 6.25amp torroids and a pair of 53,000uf 63v caps. This amp is to drive 5ohm magnepanars which are a pretty benign load, impedance stays pretty much a flat line at 5ohms. Am planing on cranking the bias up as far as I can, heatsink capacity not withstanding.
Well time to pry out the old plastic and get some parts on the way. Thanks again!
Well time to pry out the old plastic and get some parts on the way. Thanks again!
DH200 Input cap advice
A great full thanks to all who have posted here and other threads with all the fantastic Hafler info. Thanks to you all I have been able to resurrect a tired DH200 with new MC power caps, all new capacitors on the main PCB’s, upgraded internal power and signal wiring, upgraded RCA’s and binding posts complimented with a dead quiet grounding set up. I was able to replace the input transistors with new matched pairs to bring the DC offset to less than 5 mv per channel and set the bias back to factory. For a relative nuub, I’m very happy with the results!
I was wondering if I could possibly get some further advice.
From reading thru the Hafler threads, I understand that some additional performance may be achieved by replacing the input capacitor (C1) with a metallized polypropylene film capacitor. Values I have read about are in the 2 to 8 mfd range – I am curious as to the most appropriate value to install here. I have come across some Solen Fastcaps PA (250V) series with values ranging from 1.1mfd to 7.5 mfd... Slow experimentation, while evaluating performance is definitely not my forte.
Could the group recommend a starting point or am I trying to put too fine of an edge on this? Thanks in advance!
A great full thanks to all who have posted here and other threads with all the fantastic Hafler info. Thanks to you all I have been able to resurrect a tired DH200 with new MC power caps, all new capacitors on the main PCB’s, upgraded internal power and signal wiring, upgraded RCA’s and binding posts complimented with a dead quiet grounding set up. I was able to replace the input transistors with new matched pairs to bring the DC offset to less than 5 mv per channel and set the bias back to factory. For a relative nuub, I’m very happy with the results!
I was wondering if I could possibly get some further advice.
From reading thru the Hafler threads, I understand that some additional performance may be achieved by replacing the input capacitor (C1) with a metallized polypropylene film capacitor. Values I have read about are in the 2 to 8 mfd range – I am curious as to the most appropriate value to install here. I have come across some Solen Fastcaps PA (250V) series with values ranging from 1.1mfd to 7.5 mfd... Slow experimentation, while evaluating performance is definitely not my forte.
Could the group recommend a starting point or am I trying to put too fine of an edge on this? Thanks in advance!