I have a post in Solid State about wanting to get the most out of my NAD M3 that is currently in need of repair, but my thread isn’t seeing any action, so I thought I’d ask a simpler question here. As im going through the repair I thought I’d try to improve on it. I’ve seen multiple times where people say that the 2SA1943 and 2SC5200 while good, aren’t the greatest. I’ve also seen where folks have said that the MJL1302 and MJL3281 are incredible devices.
If I were to simply pull out the 2SA1943 and 2SC5200 devices and swap them with the 1302/3281 devices would that give any audible improvement? The drivers are 2SC4793 and 2SA1837. Could I improve on those?
But to start off, I’d first like to know if by changing the outputs to the better devices would I be able to hear any improvement or a change even.
Thank you,
Dan
If I were to simply pull out the 2SA1943 and 2SC5200 devices and swap them with the 1302/3281 devices would that give any audible improvement? The drivers are 2SC4793 and 2SA1837. Could I improve on those?
But to start off, I’d first like to know if by changing the outputs to the better devices would I be able to hear any improvement or a change even.
Thank you,
Dan
Hardly so. The amplifier has feedback flattening any difference. You might 'hear' things because you know you have swapped components.
Have it repaired to a working state first, the designers have choosen the components, topology and actual realisation for good reasons, both technical and commercial. If you ask such a question, your not on their level yet.
Have it repaired to a working state first, the designers have choosen the components, topology and actual realisation for good reasons, both technical and commercial. If you ask such a question, your not on their level yet.
Heat sink doesn't ever permit it in these NADs.
The MJL3281’s are a more rugged, larger die device - that’s all. The 0302/0281 are a more “direct” replacement for C5200/A1943.
The MJL3281’s are a more rugged, larger die device - that’s all. The 0302/0281 are a more “direct” replacement for C5200/A1943.
The reason to "upgrade" transistors is that the original devices failed. Improving the sounds AND improving reliability requires updating the circuit with modern practices like driver cross-coupling, LTP degen resistors, current mirrors and two pole compensation, and adding output protection if there was none before. As Self pointed out, more idle bias may make it worse and not better, and it will certainly compromise reliability. In some cases, changing the ground wiring will solve poorly designed grounding. If these things are not known to you, then you should be happy if you succeed to get it working again.
Best chances for upgrade is in electrolytic capacitors.
Edit: when higher bias is not feasible
Edit: when higher bias is not feasible
Last edited:
Actually I think the designers merely made suggestions on which components to use and the bean counters were the people that actually chose what to use. That’s why they went with the most incredible “JH” electrolytic capacitors.
Not even 20 years old yet and all have esr through the roof.
So why would you buy Onsemi MJL3281/1302 when you could get the Toshiba 2SA1943/2SC5200 for literally 1/3 the cost? When you’re using 16 output devices that adds up, selling 5000 units that’s 80,000 devices at 1/3rd the cost. If the designers and engineers had the option and cost of materials wasn’t a consideration do you think they’d choose the MJL devices? I sure as hell don’t think they’d pick “JH” capacitors by choice.
The heatsinks in this amp are pretty stout, each weighing in at just under 6 pounds each, about 2.65 kg.
And made of two pieces 180mm long, so 360mm in total.
Now seeing the heatsink, do you still think higher bias isn’t feasible? Higher bias may be better dealt with from the 3281/1302 or 4281/4302?
Wg_ski, so you feel that the MJL devices are basically the same as the 5200/1943 pair, just more robust? What got this idea into my head was what you said earlier in a thead where I was looking for MT-200 replacements for a Yamaha amp and settled on using MJL3281 and MJL1302.
“MJL2119x is quite a different animal from MJL3281. The latter is a wide bandwidth high linearity type and will usually work where Sanken LAPT types were. Maybe not a “good” as the originals but better than the TTC5200 would. The 5200 was a step BACKWARD, but it was cheap in its day. There are no more REAL MT200 types anymore - Sanken discontinued them, and everybody else’s are copies which range from “maybe ok” to “don’t even try“ bad. On’s version of the 3281 is a bigger die than the C5200 or even the original Toshiba 3281, but other than that is similar. Not exact for sure, but still a >20 MHz fT and flat beta out to 8A.”
“That’s what The Big Deal with the MT200 was - getting the heat out of the case. The better job you do there, the more the solution will resemble the original. Biggest you can get nowadays is the TO264.
Sanken is still making some of the same dies they used in the MT200’s - just putting them in the tiny case.
On Semi’s C3281 is about the same size die as Sanken used in their bigger devices. Its bigger than Toshiba’s original C3281 and the stripped down cheapo C5200. 200W vs. 150, 11 nF Cbe vs. 7.”
From those remarks I took it as the Toshibas being a fairly inferior device compared to the Onsemi. I didn’t know if that translated into the sound quality of the output, possibly lower noise or distortion at certain output levels. The MJL being higher power devices wouldn’t be run close to their limits as the Toshiba devices which again led my thoughts be to being cleaner overall. As far as I know the original outputs are run within limits, I’ve repaired two others of this model and every time it’s just bad capacitors, outputs are always fine. Then I thought “hey, I have several of these Onsemi devices, wonder if they’d give me a better sound?!?” Sounds like probably not? I’d have better luck changing the opamps?
I’d certainly hope not, they all came from Mouser or directly from Onsemi.
I have no doubts I can get it working again, I’ve repaired this exact model in the past a couple of times not to mention hundreds of amps in the past. There are far more reasons to upgrade transistors other than failure, it’s silly for you to say so, maybe you don’t know? I’ve swapped signal transistors in an amp and input sections of preamps in order to gain a lower noise floor as one example, there are several reasons to change out transistors for reasons other than “failure”. I guess when I succeed to get it work, yes I will be happy, but not surprised.
Thank you, that is indeed a much needed thing. They used absolute garbage in this amp, so I’ve been using Elna Silmic II and Nichicon FG, then Panasonic FC in certain places of the circuit not in the audio path.
Another thought I had was to replace all of the grey films (not sure of their brand or dielectric) with Wima polypropylene, do you think that would be a worthwhile effort?
Thank you to everyone that has been helpful, I’d love to hear any suggestions as to where you think I would get the most from upgrades. It’s very clear that the financial department over at NAD had a heavy hand in picking parts for this unit. It’s a very nice unit with cheap parts that I think could be greater without someone hanging from the purse strings.
Dan
Handling higher bias is a bulk heat issue. If your supply is up around +/-75V, you might be able to tolerate 100 mA of total bias. Push it to double, and even those heat sinks will get hot. Really hot when cranked and really hot just idling are not the same thing, either.
The 2SC5200 is a CHEAPER version of the 2SC3281. Less SOA at higher voltage (>50V). The TTC5200 is yet still CHEAPER, as the die is shrunk. The ON MJL3281 is a bigger more robust version, capable of what the old MT200’s were. If the C5200’s aren’t blown, they are good enough. You won’t get some magical increase in sound quality going bigger. They just tolerate higher peak power/current with more safety margin.
At one time, the price differences were negligible. Not for penny pinching bean counters, but the price differences for us weren’t worth haggling over. Those days are gone, probably for good. I have about 200 each of the 2119x’s and maybe the same # of Toshiba, between the older, newer and ON versions. All kinds of assorted TO-3’s both old and new vintage. Most of them purchased when they were under two bucks apiece. When they’re gone, that’s it. No more solid state amp building. It’s likely a lifetime supply anyway. At the current $6 each average price they’re pretty much non-starters.
The 2SC5200 is a CHEAPER version of the 2SC3281. Less SOA at higher voltage (>50V). The TTC5200 is yet still CHEAPER, as the die is shrunk. The ON MJL3281 is a bigger more robust version, capable of what the old MT200’s were. If the C5200’s aren’t blown, they are good enough. You won’t get some magical increase in sound quality going bigger. They just tolerate higher peak power/current with more safety margin.
At one time, the price differences were negligible. Not for penny pinching bean counters, but the price differences for us weren’t worth haggling over. Those days are gone, probably for good. I have about 200 each of the 2119x’s and maybe the same # of Toshiba, between the older, newer and ON versions. All kinds of assorted TO-3’s both old and new vintage. Most of them purchased when they were under two bucks apiece. When they’re gone, that’s it. No more solid state amp building. It’s likely a lifetime supply anyway. At the current $6 each average price they’re pretty much non-starters.
Dan,
I never realised NAD made such amplifiers. This is a very different level and I may assume the designers have made choises in correspondance with aimed performance, customers and price.
To answer your initial question (pull out the 2SA1943 and 2SC5200 devices and swap them with the 1302/3281), from an electronic engineer persepective only, I'd say to calculate all values of involved components thouroughly before engaging on such an adventure. Given the other thoughts mentioned above, use your second M3 as a guinea pig and compare them in a double blind test.
Thank you so much for explaining it to me. That answers that, I’ll keep the originals in and even though I already have them it’ll save me money as I can use them elsewhere. Once I power it up I’ll leave bias at factory position and see how it does and monitor the temp on the heatsinks. At that point I’ll play around with bias points.
Yeah I miss the days of cheaper parts. I try to buy all Onsemi, mostly TO-3, but like you I still have about 250 each of the MJ21195G and MJ21196G, about 150 each MJ21193G and MJ21194G and then maybe around 200 random vintage TO-3 devices that were pulls from restorations. And then maybe 4 dozen each of the common TO-264-3 case types were talking about and the same for TO-3P-3 case type, so MJL and NJW 2119x’s and such. Only ones I don’t have are the MJL1402/4281, need to grab me some of those. But, yeah I’ve gone through several, but when I originally bought the MJ2119x’s I had about 300 of each and I want to say I paid between $2.50 and $3.50 each as I bought in quantity, but I didn’t buy all 1200 at one time lol I just eventually bought up to that number. I did the same for all of the smaller transistors commonly used but in the thousands as I hope that I never have to buy parts again, at least these parts. I can’t believe that these TO-3 parts are nearing $10 each. I have to put all new devices into my PL dual 500, that’s 36 MJ2119X, that would hurt if I had to buy them at today’s prices.
Thank you for the response, unfortunately I don’t have the other M3s, they went back to their owners. They belonged to some folks I know that wanted them repaired and just like this one it was strictly caps with high esr. I run into soooo much NAD gear with bad capacitors. In the end as you may have already read, I’ll stick with what is in there since I really wouldn’t gain anything sonically.
Could very likely be! I bought this unit from a guy in a dark alley of downtown NYC, said he had an extra after a customer install and that his boss would be angry with him if he didn’t get rid of this extra unit.
Since the outputs are settled may I ask you folks for your opinion on two other things. Since I’m going through and replacing all of the electrolytics, would there be any benefit to replacing all of the grey films everywhere with Wima polypropylenes? For the gray films I don’t know what brand or dielectric they are, so the Wima polypropylenes could very likely be an upgrade, though again sonically how much? Would I be wasting time and money, or would it definitely be worth the effort? Unfortunately I cannot do a side-by-side comparison, so I am going by the wisdom of the forum.
Secondly are replacing the opamps. It uses TL082CP, I know that there are some Burr Brown devices and some quite expensive devices I could use. I was hoping to get some opinions, maybe some of you have had good luck with certain devices. I would like to keep each op amp to around $20 max, unless it’s a really really nice one and worth it’s $50 price tag lol.
If I cannot gain anything sonically with the outputs hopefully I can get it elsewhere. I am not disappointed by the sound of the M3, it actually sounds very nice, but just hoping I can squeeze more out of it since the financial department isn’t limiting me as it did when this was made.
Thank you,
Dan
I may be totally off base here since I don’t have a schematic. But my GUESS is that TL082 is a DC servo. Why? It’s sitting next to some large value film caps and a 1 meg resistor, which make a fractional-Hz low pass filter. If that’s what it’s doing it is already the right type for the job, and changing it out for a higher precision expensive op amp is useless. The op amp has nothing to do with the sound. The critical parameter is DC input bias current - and that TL082 is about as low as you can get. And changing out the caps wouldn’t buy anything either, since they don’t develop ANY AC voltage. The only thing you don’t want there are electrolytics since any leakage current at all defeats the purpose of the servo.
Ive seen (and built) many amps with TL082-ish actual input stages, but anything with those wouldn’t bother with those class A line driver modules which cost a lot more than the pennies saved on op amps.
Ive seen (and built) many amps with TL082-ish actual input stages, but anything with those wouldn’t bother with those class A line driver modules which cost a lot more than the pennies saved on op amps.
Put the bigger transistors in one channel and then compare and test it against the 5200 1943 channel.