Luxman LV103u replacing output FET's

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Dear friends,

I am currently repairing a Luxman LV-103 amplifier with shorted output FETS, so I am unfortunatelly forced to replace them along with some other broken conponents (emitter resistors etc.). But there is a small problem; the 2SJ115 / 2SK405 are almost nowere available, and the ones that are available are most probably fake.

So I sought to a nice replacement and I think I have found them; a pair of 2SJ200 / 2SK1529. Is this a proper replacement? I saw that there were some spec deviations between the 2SJ115 / 2SK405 and 2SJ200 / 2SK1529, namely different gate-source capacities and the idling current should be somewhat differend. Is this going to affect the soundquality? I was also wondering if I have to change the drivercircuit and how much lower or higher the idling current has to be? And I would like to ask if I need to match the FETS.
 
Disabled Account
Joined 2006
From what I can tell from the datasheet is that the 2SJ115 / 2SK405 are lateral mosfets so the answer is no, the 2SJ200 / 2SK1529 are not direct replacements. The latter are vertical mosfets.

I dont have schematics of that amp but if it happens to have a VBE multiplier it may be possible to modify it so 2SJ200 / 2SK1529 can be used.

A more suiteable replacement would be 2sj162 / 2sk1058.
 
From what I can tell from the datasheet is that the 2SJ115 / 2SK405 are lateral mosfets so the answer is no, the 2SJ200 / 2SK1529 are not direct replacements. The latter are vertical mosfets.

I dont have schematics of that amp but if it happens to have a VBE multiplier it may be possible to modify it so 2SJ200 / 2SK1529 can be used.

A more suiteable replacement would be 2sj162 / 2sk1058.

DEAD WRONG and you should have searched - this has been discussed at length during the years.
J115/K405 are NOT lateral, they are so called PI-channel devices which behave like vertical MOSFETs but have a low treshold voltage of about 1V or so.
J200/K1529 are their direct descendants, foa slightly higher Id max, but otherwise they can be used without changes in the LV103.
Even the J201/K1530 can be used there, regardless of slightly higher input capacitance because their increased transconductance effectively bootstraps this capacitance to the level of the originals.

Using laterals in these amps is possible but requires changes to the bias circuit AND changes to the PCB layout because lateral MOSFETs have a different pinout. Also using VMOS/HEXFET parts is possible but again, a completely different mod to the bias circuit is needed. Laterals will actually reduce the performance of this amplifier, so it would be a lot of effort to get a inferior end result.
 
Disabled Account
Joined 2006
DEAD WRONG and you should have searched - this has been discussed at length during the years.
J115/K405 are NOT lateral, they are so called PI-channel devices which behave like vertical MOSFETs but have a low treshold voltage of about 1V or so.
J200/K1529 are their direct descendants, foa slightly higher Id max, but otherwise they can be used without changes in the LV103.
Even the J201/K1530 can be used there, regardless of slightly higher input capacitance because their increased transconductance effectively bootstraps this capacitance to the level of the originals.

Using laterals in these amps is possible but requires changes to the bias circuit AND changes to the PCB layout because lateral MOSFETs have a different pinout. Also using VMOS/HEXFET parts is possible but again, a completely different mod to the bias circuit is needed. Laterals will actually reduce the performance of this amplifier, so it would be a lot of effort to get a inferior end result.

Maybe, Ill see after I ve gotten hold of proper datasheets.
 
Hi Homemodder and Ilimzn,

Thanks for your replies, I really appreciate that. So if I understand correct the FETS are suitable, and I dont have to change the idling current (bias) which corresponded to 65mv for the original devices (measured at the 0,22 ohm resistors).

Thats interesting, so a higher transconductance brings the capacitance a sort of down. Thats great to hear :) Could you maybe explain how this precise works? (I am pretty new to mosfets and their terminology).

I wanted first to replace the FETS by a pair of IRFP240 / irfp9240, but I was scared this would influence the soundquality badly.
 
Last edited:
IRFP240/9240 are not very good electrical complements (transconductance is different by a factor of over 1.5), they are only process complements (N and P doping reversed, but electron mobility is different in N and P so you end up with a non-complementary part). They also have a much higher treshold (over 3V typically) so the biassing arrangement would have to be changed. You will not be able to reach the required treshold to get more than a fraction of a mA of idle current without modification.

J200/K1529 are almost direct replacements. The output stage of the LV103 (and it's cousins) is a follower, so the larger Cgs capacitance of the MOSFET is 'bootstrapped', because the source follows the gate. Because the transconductance (a form of gain) is larger by the same factor as the maximum current with respect to the original MOSFETs, less Vgs is needed for the same current to the speaker, compared to the original MOSFETs. So, even though the capacitance is larger, it has to be charged to a smaller voltage to perform like the originals, making the larger capacitance of the replacements 'look' smaller - about the same as the originals. However, there is also Cdg, which indeed is larger by about 20%. However, Cdg is only a small fraction of Cgs, so these extra tens of pF are no problem for the driver stage.
An example of how this works is looking at paralleling the same kind of MOSFETs. Suppose Cdg is 100pF and Cgs is 1000pF, and transconductance is 2A/V - meaning, Vgs of 1V means the MOSFET becomes a 2A current source.
If you want 6A through your load, you would need to charge Cgs with 3V, for a single MOSFET. Also, you would need to change the voltage on Cds by the same voltage as 6A makes on the load.
Now, immagine you paralleled 2 same MOSFETs to do the same work - this would be like putting in one MOSFET that can handle 2x the maximum current, because MOSFETs are internally cell-based and larger ones are simply made by paralleling more cells.
Now, to get 6A, it means you need 3A per MOSFET. To do this, you now need to charge Cgs to 1.5V instead of 3V. If you had limited current to charge the capacitances, now it charges half as slow, because the total capacitance of paralleled MOSFETs is now 2x 1000pF, but needs to be only charged to half the voltage, so the net result is, as far as Cgs is concerned, the same.
However, the combined Cdg is also 2x 100pF, so if you are looking at total charge, you have an extra 100pF Cdg to charge. Fortunately, in the case of the LV103, the difference is not double but about 25%, which actually easily falls inside the tolerance margins of the design.

J200/K1529 are actually 'the next size up' J115/K405, and as close as you can get with respect to spec, tresholds, thermal characteristics etc. They are slightly larger (10A Idmax vs 8A) so offer a small measure of extra robustness. I have worked on several of the LV hybrid amp series, and these, as well as J201/K1530 are really the best and IMHO only option to use, in the absence of the originals.

It should be noted that the originals are out of production for close to 20 years now. There are small numbers around for a high price, but I have also seen fakes. IF you want to keep your LV103 absolutely original, the J115/K405 CAN be found but they will cost a lot and you may end up with fakes. Or, you can easily get genuine J200/K1529 and have an amp that will perform at least as well as the original setup, but it will also be more robust and reliable.

While you are under the hood of the LV-103, it would be wise to check all the soldering on the driver board and especially on the tube sockets (both are on the front panel on and behind the tubes). The best practice here is to remove the old solder and flux residue (you might be dissapointed by obvious thermal damage to the PCB :( ), and resolder with lead-free solder. The reason for this is thermal stress there, lead-free solder has a higher melting temperature and is less prone to forming cold/dry joints under thermal stress, but it's also less easily solderable as such.
 
Ilimzn: thank you thank you thank you!!! Aha so that is how it works, I've learned a lot today :) There are thus a lot of parameters which influence the behaviour of the FET. I think the best option for me is to go for the J200/K1529 FETS, these are well available here in Holland and not too expensive (about 10 euro's per pair).

You're absolutely right about that Ilimzn, there were a lot of bad solder joints. There was also another problem: the connectors from the driver board to the tube board were totally dried out. They had been literally crumbled to little pieces, so I had to solder new wires which replaced the connectors directly to the driverboard. I think the heat of the tubes played a role in this proces. To be fair dissapoints the quality of this Luxman me a bit, it's not the quality I am used to. Brands like Denon or Rotel are in that aspect much better over the years, while this amplifier was, if I am right, pretty expensive in the day.

Thats a clever tip, I've never thought of using lead free solder! I will certainly resolder the tubesocket contacts with lead free solder to the PCB :)
 
perhaps the alfets from Semelab/Magnatec are a good choice - ask there:
Contact -TT electronics Semelab
Magnatec. ALFET Lateral MOSFETs

These are close to original Hitachi/Renesas parts, and not directly suitable. They can be used and the mod is fairly simple but due to the peculiarities of the LV103 design you get less output swing unless you use the big multiple-die parts, which tend to be well on the expensive side - also AFAIK the different pinout problem remains.
 
Ilimzn: thank you thank you thank you!!! Aha so that is how it works, I've learned a lot today :) There are thus a lot of parameters which influence the behaviour of the FET. I think the best option for me is to go for the J200/K1529 FETS, these are well available here in Holland and not too expensive (about 10 euro's per pair).

You're absolutely right about that Ilimzn, there were a lot of bad solder joints. There was also another problem: the connectors from the driver board to the tube board were totally dried out. They had been literally crumbled to little pieces, so I had to solder new wires which replaced the connectors directly to the driverboard. I think the heat of the tubes played a role in this proces. To be fair dissapoints the quality of this Luxman me a bit, it's not the quality I am used to. Brands like Denon or Rotel are in that aspect much better over the years, while this amplifier was, if I am right, pretty expensive in the day.

Thats a clever tip, I've never thought of using lead free solder! I will certainly resolder the tubesocket contacts with lead free solder to the PCB :)

I've never had the connectors actually disintegrate :( but unfortunately, the design of many Luxman units of that era (even much more expensive ones) shows a distinct lack of concern for thermal problems, mostly to do with using cheap FR2 PCB material and lack of heatsinks combined with dense spacing of heating parts. They would operate to spec well after the waranty runs out but it's not the longevity one would see in previous late 70s units... cutting costs and insuring repeat customers after a few years when repairs would turn out more expensive than replacing the unit? Rather risky strategy if you ask me...
 
Hi,

I have some unused 2SK405/2SJ115 hanging around. I bought them more than 20 years ago in Japan (fakes weren't around at that time, old good days). I used them in my previous amps and these are left over from those projects. As I don't have any plan to use them soon, I'm OK to part with them...

As ilimzn wrote, successor of 2SK405/2SJ115 pair was 2SK1529/2SJ200 pair but the difference in transconductance and input capacitance may lead to a shift in poles, which is a slight concern.

Good luck!
Satoru
 
I've never had the connectors actually disintegrate :( but unfortunately, the design of many Luxman units of that era (even much more expensive ones) shows a distinct lack of concern for thermal problems, mostly to do with using cheap FR2 PCB material and lack of heatsinks combined with dense spacing of heating parts. They would operate to spec well after the waranty runs out but it's not the longevity one would see in previous late 70s units... cutting costs and insuring repeat customers after a few years when repairs would turn out more expensive than replacing the unit? Rather risky strategy if you ask me...
I agree. Not only by Luxman. Also by brands like Krell and Mark Levinson. The No-23 and the KBL are examples.

The main reason for this situation was always very highly desired power output (sometimes also the voltage output by preamps) for advertisements and brochures, and thus too high supply voltage for the /preamp/predriver stages and driver stages of power amp sections (hot spots and burned PCB aeras and damaged electrolytic caps was mostly to observe as consequence).

In such cases only one way exist for repair work to reach a perfect solution: Design of an outdoor power supply with only half of the supply voltage for the power amp section so as creating an appropriate interface and redesign at certainly places therefore.

This means a quarter of power output (which is far less than it sounds - only -6db maximum undistorted audible power output).
It continues to give very many design options like quality standarts for the caps and transformers (space is no longer an issue) and many more.
For the output power stage now one can realize higher idle current values for better sound (in direction to pure class A).

OTOH - the retention of the original/genuine state is often much more easier (no additional power supply components) but in all cases always a big compromise in respect to the realibility and sound quality.

Dear friends,

I am currently repairing a Luxman LV-103 amplifier with shorted output FETS, so I am unfortunatelly forced to replace them along with some other broken conponents (emitter resistors etc.). But there is a small problem; the 2SJ115 / 2SK405 are almost nowere available, and the ones that are available are most probably fake.

So I sought to a nice replacement and I think I have found them; a pair of 2SJ200 / 2SK1529. Is this a proper replacement? I saw that there were some spec deviations between the 2SJ115 / 2SK405 and 2SJ200 / 2SK1529, namely different gate-source capacities and the idling current should be somewhat differend. Is this going to affect the soundquality? I was also wondering if I have to change the drivercircuit and how much lower or higher the idling current has to be? And I would like to ask if I need to match the FETS.

BTW - what was the actually reason for the damaged MOSFET's from the output buffer stage in your case?
If you don't know this exactly, you cannot carried out a replacement without the risk of another immediate (and expensive) destruction.

Here an audio MOSFET overview - perhaps this helps a little:
http://www.diyaudio.com/forums/soli...l-mosfets-class-ab-overview-p-spice-urls.html
 
Last edited:
It was a bit surprising for myself too Ilimzn, I had never thought the connectors could desintegrate so bad :-( I guess this amp is used quite a lot. Fortunatelly the tubes look pretty new to me, they perform also pretty good.

Wow I did not know even Krell had problems with bad thermal design, Tiefbassuerbertr. I am reproducing a KSA-100 (an original KSA-100 is a bit too expensive haha) as a sideproject. It's a big challenge to keep this beast cool.

I think the FETS in the LV-103 are also pushed a bit to far to their SOA's, I've red that a lot of this Luxmans have problems with broken FETS.

Well tiefbassuerbertr, I've bought the Luxman from a guy who tried to repair the Luxman but didn't succeed. He had soldred a transistor wrong on the driverboard (2SA1361) and there was a problem with the tubeconnectors. The driverboard didn't make contact with the tubes due to the desintegrated connectors. So I repaired this connections with 1 mm silvercoated wire instead of connectors. So far so good, and the amp was performing well again. Last thing I wanted to do was changing the bias to the correct value (bias was slightly low), but that went terribly wrong. I was plugging a chinch connector in the amplifier at the front (AV) when suddenly one of the measuring clips got loose and made contact with the heatsink. So there was a direct short circuit from the emitter resistor to the heatsink, result one of the channels was blown, next to two fusible 100 ohm resistors and the emitterresistor :-( I felt so stupid! Almost the whole clip is isolated, there is only 2mm blank metal on this clip, and exactly this part made contact with the heatsink :-( But sh*t happends they say.

I've continued yesterday with measuring the driverboard and found also a 2SA1361 with strange values when I tested it with the diode test. I got a value of +/- 0,400mv while this had to be around 0,700mv. So I am pretty sure this one had to be replaced as well. Unfortunatelly the 2SA1361 is almost not available in Holland, so I've checked the VRT ECA database and found I could replace this one with 2SA1479..80, (BF 416, BF 418, BF 470,++). The BF470 is well available in Holland, but is this a suitable replacement or is it better to stick with the original design?

Thanks for the overview tiefbassuerbertr!!! Thats a handy topic! I really appreciate this forum, there is so much knowledge around here :)
 
Since the design uses BJT drivers, and it's a follower, fortunately different capacitances and gm change the position of the poles, shich could be an issue for compensation, slightly. Production amps are almost as a rule slightly (and sometimes grossly!) overcompensated to alow for component tolerances and often also for replacements, in case a part stops being produced or available during production. The J200/K1529 and also J201/K1530 have neen tried and tested by many in the LV10x range, without problems. That being said, different transconductance also requires a slightly different adjustment of the idle current, unfortunately as with all MOSFETs, you can't really make recomendations. They should all ideally run in class A for best results :) but obviously real world limits such as power supplies and thermal management force a compromise. To be honest, I don't regard these amps as one of Lux's better constructions, one thing where they certainly could have done better id the tube choice (not that there's anything wrong with the tube, there are others more suited to the working conditions at hand). In any case, I've always adjusted the ones I repaired / refurbished looking at a distortion meter, or, later, spectrum analyzer.

The FETs indeed are pushed far, especially in the LV105, same fets, higher voltage, more power. They are much more fragile than laterals, and the protection is minimal.
I have to say, though, none of the units I fixed with J200/K1529 or J201/K1530 ever came back with that problem, but it could be just because their users were more careful after blowing up the originals...
 
Last edited:
I had repaired my LV103 by replacing busted filament transformer and output Fets. I had replaced them with laterals k1058 and J162 and moding the bias.at this time they are good in measurement but wharen i plug in music it distorts as volume increase then go in protection mode. I was dissapointed, then I tried replacing it with Irf9240 and Irf240, and revising resistor values in the VbE multiplier, after adjusting bias on the other channel and played music its damn good. but when the other channel is tested is its ocsillates and heat up Fets then burned. I tried putting gate stopper from 470 to 1.5k ohm but still it oscillates...tried to replaced driver xtrs fbut still the same.....anybody here knows what to do with this oscillations on one channel only...

I'm using the other channel right now to drive my center speaker but i want the other to work also...
 
Haha, maybe my FETS went for a few uS into class A before they blew ;-). Hey thats interesting, I was wondering if the tubes were really that good. I think the ECC83 would perform a bit better, but I think the anode voltage has to be increased a bit to put them in their lineair area. Same for the original 6CG7 which is currently in the Luxman. I measure about 100V on the anode. I think thats slightly low because Va would ideally be around 250V, or am I thinking the wrong way :)? By the way, does anybody know what the capacity of the 4 small (ceramic / mica?) capacitors on the tubeboard is? I would like to replace them aswell because they are turned to black due of the heat from the tubes. Below a picture of them (and also the disintegrated connectors):

An externally hosted image should be here but it was not working when we last tested it.


Thats bad news Junm :-( Oscillations are deadly for the output FETS. Why don't you try the J200/K1529? A lot of people had great results when using them.

By the way, I've replaced the FETS with J200/K1529. But I have to fix the driver stage before I can test if the operation went OK. I've also maybe bought the original FETS from Satoru :)

I have unfortunatelly not a distortion meter or spectrumanalyser :-( Is there a way to measure the distortion with a scope? Off course you can see if a sine wave is undistorted, but are there other methods? Maybe it's a good thing if I invest some money in a distortion meter, I think this would be a handy apparatus for restoring amplifiers.

Well here are the culprits, the broken FETS:
An externally hosted image should be here but it was not working when we last tested it.


And the new ones:
An externally hosted image should be here but it was not working when we last tested it.
 
Here several threads from far east with good pictures:
ALPINE/LUXMAN LV-103
Amly Luxman Lv-103u - Hà N?i | Amplifier | Mua ngay, m?c c? | Shop lamaudio
Shop : Dainhan_audio | Âm thanh hifi, Karaoke
ALPINE/LUXMAN LV-103
????????????????????????????? - ????????????????????? - Yahoo!???
??????????? - Yahoo!???
I have download the schematic diagram about
Luxman manuals, schematics and brochures - Hifi Manuals
and I note, that one must redraw the power amp section for quickly understanding (too much and too long drawed wires - as usual by japanese diagrams)
Because the schema isn't an ordinary topology, redrawing according good understanding from the therory of operation and finding possibly existing design errors is a good help.
The kind of bias servo (Q7307/Q7309) I haven't see before - go to
http://www.diyaudio.com/forums/pass...as-automatic-self-biased-overview-wanted.html
BTW obviously there are various versions of the heatsink - last image with horizontal mounting surface for the MOSFETs.
 

Attachments

  • Luxman LV103u power amp section, buffer.pdf
    164 KB · Views: 226
  • Luxman LV103u power amp section, bufferII.pdf
    169.9 KB · Views: 170
  • Luxman LV103u-LV105u Owner's manual cover.jpg
    Luxman LV103u-LV105u Owner's manual cover.jpg
    23.7 KB · Views: 321
  • luxman_lv_103_int1-I.jpg
    luxman_lv_103_int1-I.jpg
    95.7 KB · Views: 616
  • luxman_lv_103_tubes.jpg
    luxman_lv_103_tubes.jpg
    21.1 KB · Views: 254
  • luxman_lv_103a_internal.jpg
    luxman_lv_103a_internal.jpg
    167.8 KB · Views: 818
  • Luxman LV103u power amp section, front end.pdf
    27 KB · Views: 192
  • LUXMAN_LV-103.jpg
    LUXMAN_LV-103.jpg
    78.9 KB · Views: 254
Last edited:
Haha, maybe my FETS went for a few uS into class A before they blew ;-). Hey thats interesting, I was wondering if the tubes were really that good. I think the ECC83 would perform a bit better, but I think the anode voltage has to be increased a bit to put them in their lineair area. Same for the original 6CG7 which is currently in the Luxman. I measure about 100V on the anode. I think thats slightly low because Va would ideally be around 250V, or am I thinking the wrong way :)

ECC83 would certainly be about the WORST tube to use there. A fair amount of current is required, which the ECC83 simply isn't capable of.
But, you are right in one thing - the anode voltage is very low for a 6FQ/CG7 to get the proper current going, even though this tube (essentially a noval version of the ever popular 6SN7) is capable of great performance, in this design you cannot put the operating point into the proper place to get this performance. An ECC88/6922 and numerous subvariants would have been much better here, far better performance at lower voltages and also capable of delivering more current at said voltages. Also, has about double the gain in this sort of application. Lots of other tubes would be good candidates but not nearly as popular or readily available. In case someone gets ideas, NO you can't just drop in a 6922, mods would be required, not the least of which would be overhaul of the compensation networks, and absolutely mandatory use of grid stoppers. The 6922 is a VHF tube, quite high gm - these were made for RF work, and if you don't give it to them, they will gladly make some themselves if you give them half a chance :)
 
Is there a possibility that Vas tubes are causing oscillation on the output FETS? I'm planning to replace the Tubes with russian 6n1P or 6N3p....will this be good?

I can't avail here the k1529/30 and j200/201 and very costly too, so i Settled on the IRFp's. Sounds good on my other channel. My laterals are defective even they are brand new bought on the internet (ebay).
 
Haha whoops, so the ECC83 is not a great replacement.

I am still waiting for parts before I can finish the Luxman. I hope I will receive them next week. So this operation is to be continued... ;-)

By the way, something I am don't really sure of, can I replace the broken 2SA1361 with a BF470? It should be possible, but will this affect the sound in a negative way?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.