Have a hafler dh-220 amp with a dead right channel, DC voltage on the terminals. Pulled the power transistors (2-2sk134 and 2-2sj49's) and checked, 3 dead , 1 with leaked so plan to replace them all.
I read several posts about substitutes like Semelab 900/905, exicon etc but I have several concerns with those substitutions:
1) They are 125 watts and my channels would be uneven (only if slightly) and wasn't planning on (cost) to purchase 8 and do both channels.
2) They are 8 amps and current fuses are all 7a. Maybe a minor cost but
3) Concerned about other impacts to the circuit. I really don't want to mess with the board. I would prefer just to replace the transistors and be on my way.
When I look up the original data sheets, the 2sk134 datasheet is extremely similar for the 2sk133 and 2sk135, with the only difference I can see is the breakdown voltage (120/140/160). ** Is there any reason why I can't use either of these as a direct replacement? They seem to be more accessible (cheaper). Same story on the 2sj48/2sj50 What is the practical impact of the various breakdown voltages? Am I ever going to hit any of them?
I read several posts about substitutes like Semelab 900/905, exicon etc but I have several concerns with those substitutions:
1) They are 125 watts and my channels would be uneven (only if slightly) and wasn't planning on (cost) to purchase 8 and do both channels.
2) They are 8 amps and current fuses are all 7a. Maybe a minor cost but
3) Concerned about other impacts to the circuit. I really don't want to mess with the board. I would prefer just to replace the transistors and be on my way.
When I look up the original data sheets, the 2sk134 datasheet is extremely similar for the 2sk133 and 2sk135, with the only difference I can see is the breakdown voltage (120/140/160). ** Is there any reason why I can't use either of these as a direct replacement? They seem to be more accessible (cheaper). Same story on the 2sj48/2sj50 What is the practical impact of the various breakdown voltages? Am I ever going to hit any of them?
Hi,
usually it's not a problem to replace the original 2SK134/2SJ49 with the 2SK135/2SJ50. But there will be a nother issue, that has to be solved. AFAIK, Hafler made a preselecction of these FET's by Ugs. Can be seen, that the orignal FETs are numbered from 2 to 5, depending on the Ugs. Especially at the DH-220 and DH-200, where two Fets are in parallel configuration.
It looks to me, that there are more offers in the net for 2SK135/2SJ50, but take care of fake types.
usually it's not a problem to replace the original 2SK134/2SJ49 with the 2SK135/2SJ50. But there will be a nother issue, that has to be solved. AFAIK, Hafler made a preselecction of these FET's by Ugs. Can be seen, that the orignal FETs are numbered from 2 to 5, depending on the Ugs. Especially at the DH-220 and DH-200, where two Fets are in parallel configuration.
It looks to me, that there are more offers in the net for 2SK135/2SJ50, but take care of fake types.
Yes, 2sk135/2sj50 are excellent replacement parts but there is another issue here:
DH-220 schematic shows that there are 2 pairs of output MOSFETs per channel, connected in parallel without source resistors which means that tight matching is needed between MOSFETs of same polarity. Otherwise, sub-optimal operation is to be expected.
DH-220 schematic shows that there are 2 pairs of output MOSFETs per channel, connected in parallel without source resistors which means that tight matching is needed between MOSFETs of same polarity. Otherwise, sub-optimal operation is to be expected.
If you find it hard to source genuine TO-3 parts you could consider also using the 2SJ160/1/2 and 2SK1056/7/8 TO-3P parts, if I recall it correctly they are physically compatible with TO-3 and can be dropped directly in as the leg space, distance from one of the mounting holes is exactly the same as TO-3, just bend down the outer legs and solder it in place, and the middle leg can be connected through the other mounting hole which is left over from the TO-3.
Not familiar with ugs but you are correct the 134 are marked with 3's, the 49 with 4's (opposite on the other good channel) So I assume this is some form of matching like Hfe? Doesn't really matter whether it's a 3 or 4 apparently as long as they are the same. i doub't I'm going to get/find a matched pair, even if I had some way to determine it's ugs. So the question is, to the layperson (not a professed audiophile that can tell the different between .1% distortion and .01%), will an unmatched set really be noticable or is this something more practically only equipment will measure?
That's a thought, though how to deal with the heat sink mounting? The TO-3's are flush mounted (with the plastic wafer in between)
Ugs or Vgs (voltage between gate and source) controls the current between the drain and source of the mosfet (Id or Ids). Matching is the process that you perform in order to select parts which have same Id when same Vgs is applied.
If you parallel unmatched parts without source resistors, one of the paralleled mosfets will always conduct more current than the other and it has noticeable impact on amp's performance. You'll have to either match the mosfets or modify the circuit by introducing source resistors (which will also noticeably change the amp's performance).
Buying matched pairs from reliable source is your best option.
If you parallel unmatched parts without source resistors, one of the paralleled mosfets will always conduct more current than the other and it has noticeable impact on amp's performance. You'll have to either match the mosfets or modify the circuit by introducing source resistors (which will also noticeably change the amp's performance).
Buying matched pairs from reliable source is your best option.
To get around matching you could replace two mosfets with one ALF16N16K and the other two with one ALF16P16. If you can still find them. I know one of them is still available at RS Components.
But I think that brings me back to the original problem unmatched between left and right sides. Those ALF are 16amps, 250 watts.
Given I'm not likely to find two pairs *matched* (haven't seen anyone sourcing them that way thus far) I recognize the performance will be impacted but the question goes back to - will the layperson really be able to tell?
That' a good idea
One of those ALF16N16K/ALF16P16 is equivalent of two matched paralleled 2sk135 or 2sj50. So one pair of ALFs per channel does the job all right.
The ALFs do not make much sense, because the P-Fet is hard to find, maybe that's why the price is pretty low for th N-channel Fet (About 5 GBP)
Another idea is, using the ECF10N20-S and the ECF10P20-S from Exicon, distributed by Profusion plc, UK. They are available in TO-3 cases, and they are selected in different color bands, similar to the matching procedure of Haflers.
I still have 10 genuine Hitachi 2SJ49 (from the 80ies), but the counterpart 2 SK134 i don't have.
Good luck for your decision
I still have SJ55 and SK175 in tight matching.
20 USD /pair. Pm me if you're interested.
Regards
Figge
20 USD /pair. Pm me if you're interested.
Regards
Figge
No. The output power depends on power supply and speaker impedance.
Both channels will be the same.
So if i understand everyone - Whether I use a exicon, AFL, Sj55/SK175, sj50/sk135 etc it makes no difference in terms of equal volume matching to the other side originals sj49/sk134, whether volume is set to 1, 5 or 10, both channels will be equal and I won't need to increase any fuses or make any circuit changes (AFL exception since it's going from two pieces to one). That's an important point that I think wasn't clearly stated until now.
My meter measures HFE - is this the same is or similar to the Vgs? can it be used for matching these?
Does anyone know what the preselected Ugs number group 3 & 4 related to in terms of voltage/curent range?
Or look in our swap meet section .
http://www.diyaudio.com/forums/swap-meet/300659-fs-2sj50-2sk135-mosfets.html
http://www.diyaudio.com/forums/swap-meet/300659-fs-2sj50-2sk135-mosfets.html
Plot thickens as I learn more about these.
Because I had DC voltage at the speaker terminals I "assumed" one or more of the channel transistor had shorted. I put them on my B&K 520 transistor tester "assuming" it would identify the NPN vs PNP, shorts etc. When it came up not being able to determine conclusively which type it was (this model you hook up all three leads randomly and cycle a swicth and it identifies the various pins you are connected to) but since it showed on two setting where normal there would be only one, and on each setting showed a current leakage, I "assumed" it was bad. At least 3 of the 4.
Turns out - a totally inappropriate test for a mosfet. Now I understand the multimeter approach (e.g. building a quick charge up on the gate, seeing it on the drain, and come to find the two 2SK134's seem to be fine and it's the two SJ49 (one of which I "assumed was fine" that are bad. I did some other comparative reading with the old transistor tester with the good channel ones (both types) and now pretty convinced my problem lies with the SJ49s alone. What doesn't makes sense to me is since they don't show a short (unless it is happening at some high voltage) how I could have DC voltage out the speaker terminals?
Because I had DC voltage at the speaker terminals I "assumed" one or more of the channel transistor had shorted. I put them on my B&K 520 transistor tester "assuming" it would identify the NPN vs PNP, shorts etc. When it came up not being able to determine conclusively which type it was (this model you hook up all three leads randomly and cycle a swicth and it identifies the various pins you are connected to) but since it showed on two setting where normal there would be only one, and on each setting showed a current leakage, I "assumed" it was bad. At least 3 of the 4.
Turns out - a totally inappropriate test for a mosfet. Now I understand the multimeter approach (e.g. building a quick charge up on the gate, seeing it on the drain, and come to find the two 2SK134's seem to be fine and it's the two SJ49 (one of which I "assumed was fine" that are bad. I did some other comparative reading with the old transistor tester with the good channel ones (both types) and now pretty convinced my problem lies with the SJ49s alone. What doesn't makes sense to me is since they don't show a short (unless it is happening at some high voltage) how I could have DC voltage out the speaker terminals?
- Status
- Not open for further replies.