Some thoughts on Elekit TU-8200DX
Last year I put together one of these amps mostly for headphone use, based on the good to rave reviews from all forums. I did not exactly got the sound that I had expected. The highs were sweat and nicely rounded but overall it sounded somewhat muddy, laid back, lacked “PRaT” and involvement in general. I used my old trusted Audio Analogue Paganini CD and Sennheiser HD600, and the impression was certainly not made by either of the source or the transducers. I also heard some hum without signal in UL and pentode modes as well as some RF interference occasionally.
What I am looking for is not some AM radio sound from the 50’s but realistic sound, clarity, imaging and the low level details that you expect from a single ended tube amp. Replacing the stock Chinese 12AU7 driver tubes with Tungsram ECC82 led to a significant improvement in clarity, but did not change the overall sound characteristics of the amp.
Being an electrical engineer and an audio enthusiast myself (I designed and built solid state amps over 25 years ago), I analysed the circuit that is available in the manual. The conclusion is that it has great potential especially as a headphone amp but has some area where it could be further improved.
- The best feature is the FET filtered (not stabilised, filtered) power supply but I suggest a few changes in values. The 180 µF (C22) smoothing capacitor is just barely adequate especially with a 50 Hz power line that I have here, because the remaining ripple voltage drops the source-drain voltage on the FET close to the limit where it can operate at all.
- The output power specifications are a bit too optimistic. With acceptable levels of distortion you can get about 7 Watts per channel in pentode mode, 6 W in UL mode and 1.4 W (!) in triode mode from a 6L6GC. Note that it is not entirely impossible to reach to 4 W with these tubes in triode mode, but you need much higher plate voltage, lower bias current, and higher load impedance for that.
- The heating voltages are on a low side because the power transformer has a 6 V winding instead of the standard 6.3 V. This will drop further when using output tubes with high heater current requirements like EL34 or KT88, getting close to the minimum of 5.8 V for these tubes. The DC heater voltage for the driver tubes is also low. The Point# 28 voltage range is 5.4 – 6.0 V, but anything below 5.7 V is way too low. (I measured 5.9 V with 6L6GC, which would become lower with any other tube.)
- The auto bias circuitry is basically a negative feedback from the output to the cathode of the power tubes via a first order low-pass filter formed by R23/C13 and R24/C14. Although the cutoff frequency is low (below 1 Hz), due to the high amplification of a bipolar transistor, the low frequency signal slightly modulates the bias voltage up to about 100 Hz. It’s not a surprise that the later models use more sophisticated approaches.
- The overcurrent protection is based on the same filter hence it is slow. It may protect the power supply from old tubes that gradually lose bias stability but if they have a short then Q3 or Q4 is most likely already dead when the front LED turns into red. This happened to me once as well. The good news is that in most cases all you need to replace is the FET, if you are very unlucky then you can smoke R17 or R18 as well. I strongly suggest you buy a couple of spares especially if you plan to do power tube rolling with old tubes..
- The headphone output is very simple and fits high impedance phones only as the output impedance is almost 100 Ohms. Elekit even recommends taking one pair of resistors out to reduce hum, see Symptom# 6 on page 15. I don’t think this is a good idea as doing this will further increase the output impedance. Instead, just increase the value of C22 and it will be fine.
- The source of RF interference is the cable that connects the power switch on the front to the back of the amp (between Units 2 and 7) where the unfiltered power line voltage is passing through inside the amp. A mechanical connection would be better, but there is a simple workaround. Twist the cable around about four times between the back and front panels and it will be fine. Also, twist the cable around that connects the primary winding of the power transformer to Unit 7. This eliminated the problem.
- The RCA plugs use a thin metal plate and the connection is not reliable. I had to move the cables around in a few cases to get the signal back. If you do not need the jack input on the back ( I don’t), just remove the whole Unit 6, buy some Neutrik or similar RCA plugs, and connect them to Unit 5 with insulated wires. This will also eliminate the soldered connections between Units 5 and 6, which do not seem reliable for long term either.
Finally, the sound…I managed to redesign the driver section without having to mess up the PCB and now it sounds exactly as I prefer. It sounds so well now that I gave up on the idea of buying the TU-8600. I can share the details later if someone is interested.
Here are my comments about your feedback.
Yes, the stock 12AU7 is the problem. This is the first component I recommend to replace it.
TU-8200 is designed based on 6L6GC. No matter what output tube you want to use. 6L6GC is always the best match with TU-8200.
Mr Fujita is one of the best Tube amp designers in Japan. Tube amp design is an art work. Sometime I want him to change the design to make it sounds better in theory. He always explains to me why he wants to his way... I always accepts his explanation. As we know there is no prefect setting... i.e. trade-off between quality and price...
Good back to you question
Is TU-8200 better than TU-8600?
TU-8600 is the first amp kit formally reviewed by Stereophile. The review should be available in early Feb/Mar edition. This will tell you a lot....
Is TU-8600 better? Just TU-8600's stock OPT is 2.5 times heaver and bigger than TU-8200's OPT..
There is a big gap between TU-8600 and TU-8200. How big.. there are a lot of unbiased comments available in the web.
Thank you for your feedback. I was sure you would send a few comments :)
I did not mean to say TU-8600 is not a great amp or TU-8200 would be better.
What I was saying is that to me the 8200 with modified driver stage sounds good enough as a headphone amp as long as the phones have >200 Ohms impedance. For more sensitive or low impedance phones some reconfiguration of the headphone output may be necessary as it is not user configurable unlike the 8600 that offers a few options. As a speaker amp, the 8600 is almost certainly better at least at relatively high volumes thanks to the bigger OPTs not to mention the Lundahl upgrade option.
I had read most available 8200 reviews before I bought the 8200, which all were very positive with the stock tubes even though we both agree that those driver tubes are not all that great. This confirmed that these reviews need to be taken with a grain of salt just like any equipment review in the magazines or in the net. This is a general rule and has nothing to do with Elekit.
I know the 8600 also got great reviews but I could not find a review that would compare these amps under the same conditions. Those who made some comparisons tended to agree that the 8200 is slower and warmer sounding, which is something that can be fixed.
If I ever buy a speaker amp kit then I will need some more power than either of these SE amps offer, which is another reason why I did not order the 8600.
While building the 8200 I actually contacted Elekit asking if they would ever release a parallel SET amp. (I have some reservations about class A/B push-pull amps in general, which is why I did not order a 8340.) It was in mid 2017 when the 8600 was still in the pipeline and had not been announced yet. They mentioned that an SET amp would be released soon and that they were still working on that one. They said that a parallel SET version depends on the demand from the main markets especially in the US. (This is not a big surprise.) I hope that will be enough demand to justify the release of that model. (Well, I could accept a SET 845 instead but the voltage range these tubes operate with would probably be too risky for a commercially available kit.)
I just recently finished building a TU-8200 with the DX capacitor upgrade.
I would be very interested in the details of you modifications.
I can send you the changes in PM unless other builders are interested.
I am also interested in learning about your modifications.
I also have a TU-8200 that could use some upgrades :)
Give me some time as you will need the part numbers in some cases - the bigger capacitors have to fit the space available.
First of all, let me tell you what I do not recommend you to change.
a) I use the stock metal film resistors. Feel free to order a carbon film resistor package, but please consider that these components went obsolete back in the 60's and for good reasons. Their noise, linearity, long term stability are all inferior to a metal film resistor without any measurable benefit on the other side. I don't know if any contemporary commercial amp is available at all that is built with carbon film resistors.
b) I installed the Amtrans AMCO coupling caps from the DX kit and they are fine. I know the Mundorf EVO Silver Gold are preferred by many builders, but these caps also received mixed reviews. If there is any difference, it must be subtle anyway. Spend your budget on better tubes instead of boutique capacitors.
c) I ordered a TKD 2CP601 as it was recommended as the best potmeter for the 8200. Unfortunately I am disappointed. It is a pain to install, you need to make irreversible changes to Unit 2, find a custom spacer and connect the pot with insulated wires. It thought it would be more logarithmic than the little Alps pot (it is not), it would at least sound slightly better (the difference is negligible if any), and it would have good channel balance. I may have got a lemon, but my sample is way more out of balance at low volumes than the stock pot. I would have returned it if these issues had come out before soldering.
Try Alps RK27 or a good stepped attenuator instead. Or leave it as is.
Thank you for your latest (#8) post, all this is really valuable information.
I also considered utilizing an Alps RK27 and I also rejected the idea
for not wanting to cut the UNIT-2 PCB. It would be nice to have a separate
volume control PCB in the kit. I think the TU-8600 has that feature.
OK, so here is the first part - the changes that are related to the power supply with one exception. I do not call them "upgrades" as this word in DIY audio usually means a component with the same primary specs but supposedly better quality.
I recommend these modifications even if you are happy with the sound in your actual configuration. Here is some background information.
- I suspect Elekit wanted to keep the price of the kit under $700, and supplied the amp with electrolytic capacitors having the lowest value that is adequate for the particular job. I significantly raised the value of C22,C31,C32 for the above reasons until space restrictions were met.
- They also used standard values as much as possible, which is why you see many resistors and capacitors with the same values (180kOhms, 220μF/16V, 1μF/50V, etc.). This is perfectly fine, they do the job. The most extreme example is probably the bridge rectifiers. D5,D6,D7 are all the same even if they are used for very different purposes. D7 works the closest to its limits, and I suggest changing it to a 2A version even if you want to use 12AU7/ECC82 as the driver tube as the forward voltage drop will be a bit lower, which relates to the next topic.
- The power transformer provides 6V on its heater taps in all Elekit models that I saw. The nominal heater voltage for indirect heated tubes is usually specified at 6.3V +/- 10%, but all design books say that it should be kept within 5% to ensure long tube lifetime. I measured 5.9V AC on the power tubes and the same DC voltage on the drivers in the stock configuration, which are outside of the recommended range. If you install power tubes with higher heater current requirements then these voltages will drop further.
I contacted Elekit and asked why they use 6V. Initially they said the voltage was fine and they had not received any complaints, and that it may be my power line that is below the EU standard 230V. (I measured 234V.) Then when I wrote what changes I was planning to do, they said "Our kits have plenty room for modification...if you want to bring it to very mid range within the allowance, what you did to the amp makes perfect sense."
There isn't too much you can do with the power tubes as they are AC heated directly from the transformer, but if you install the bigger bridge rectifier and smoothing capacitors C31 and C32 then R64 will no longer be required. Its low value does not do too much ripple filtering anyway but it does decrease the DC heater voltage by a few hundred millivolts. After all the changes I had made to the heater circuit the DC voltage on the heater pins of the driver tubes increased to the nominal 6.3V from 5.9V, still in the stock tube configuration.
- There is no delay to the plate voltage after switching on the amp with cold cathodes, but the B+ gradually builds up. The higher the C23 and C24 values are, the slower the full plate voltages will be available.
- Everyone who made capacitor upgrades seems to have overlooked the importance of C13 and C14. These capacitors define the single cutoff frequency for the auto bias circuit, and the low ESR is very important in order to eliminate the feedback of the output signal to the cathode of the power tubes. It should be included in the DX upgrade package as the values are the same.
At this point I contacted Elekit again asking if another 1μF/50V capacitor connected between Points 13/14 and GND could create a 2nd order filter for better output signal filtering on the bias voltage. They said it could cause oscillations, and they are probably right. Any additional capacitor increase the phase shift with almost 90 degrees and due to the high open loop gain achieved by the bipolar transistors, it can lead to low frequency oscillations.
I decided to increase the value of C13/C14 instead as it will drop the signal feedback by 7 dB, which makes it negligible over about 50Hz. C23 and C24 need to be increased for tube protection beforehand as the cathode bias voltage will settle slightly slower with the higher C13/C14 values. The protection circuit will also react slower but it is slow already as I wrote in my first post in this thread, so this will not make that much difference.
I attached the spreadsheet that contains all the changes I made to the power supply and the bias circuit. Some of them like the slight increase to C33-C35 are not very important. I ordered parts that were easily available, you can use any other brand but check the size of the capacitors before ordering.
|All times are GMT. The time now is 06:41 PM.|
Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.
Copyright ©1999-2020 diyAudio