Jump to content

Hints on maintaining / improving my Stax SRM-T1S


Recommended Posts

The current through the bias trimpots is the sum of the currents through the two tube sections, which is determined in the stock amplifier by the plate output resistors, the positive voltage supply and the fact that the plate voltage (ideally) is 0 volts. Since the nominal positive voltage supply is +320 volts, and the summed plate output resistor is 66 kilohms, by Ohms law the current through each tube section has to be 325/66k = 4.85mA, so for two tube sections (per channel), the total current is 9.7mA. 10mA is close enough for calculation purposes.

  • Like 1
  • Thanks 1
Link to post
Share on other sites

Just a summary of what I have learned on planning the recap of my SRM-T1S unit - a bit on the TL;DR side and probably repeating a lot of boring stuff well known to experts in the forum (specially ones with little or no patience for capacitor esoterics - yeah, I did notice that the latter is a thing in the audiophile community 🙄 ), but it may be useful to others (or at least to me) in the future to have everything in one place. I still have a few doubts, though (those only interested in that may skip to the last paragraph below)... Thanks again to all who helped!

My current (stock) configuration of PS electrolytic caps is the following (cap numbering follows the schematic):

  • C9-10: 2 x 10uF, 50V (Marcon, currently United Chemi-Con), radial - dimensions (LS = lead spacing x D = diameter x L = length): 5mm x 10mm x 12.5mm
  • C11-14: 4 x 100uF, 400V, 85oC (Hitachi, currently AIC Tech), snap-in - dimensions (LS x D x L): 12.5mm x 30mm x 60mm
  • C21-22: 2 x 220uF, 10V, 85oC (Elna), radial - dimensions (LS x D x L): 3.5mm x 6.5mm x 11.5mm
  • C23-24: 2 x 47uF, 35V, 85oC (Elna), radial - dimensions (LS x D x L): 3.5mm x 6.5mm x 11.5mm

Surveying through a number of places (based on the above comments by Pars, Fitz and JimL) pointed that caps meant for power supply should privilege not only low dissipation factor (DF), but also high ripple current and long endurance (assuming capacitance and voltage rating the same) - high ripple current suppression capabilities are not relevant for signal-path caps (which assume that the DC power has already been filtered enough in the power supply stage) but are critical for power supply caps. "Audio grade" electrolytic caps are usually strong in the first characteristic but rather weak in the last two. They are also more often than not unavailable with the 105oC temperature rating recommended by Pars above (which seems reasonable since the circuit gets quite warm, as a pure class-A tube amp should). Taking that into account when searching through the Mouser catalog, it seems to me that the following substitutions should yield a good result within Mouser's current stock availability:

  • C9-10: 2 x Panasonic EEU-FR1H151B (FR series) - 150uF, 50V, 105oC, DF=0.10 (at 120 Hz, 20oC), ripple current = 0.82A (RMS, 120 Hz, 105oC), life = 6000h, dimensions (LS x D x L): 5mm x 10mm x 12.5mm - https://br.mouser.com/ProductDetail/667-EEU-FR1H151B
  • C11-14: 4 x EPCOS / TDK B43547A9477M000 (B43547 series) - 470uF, 400V, 105oC, DF=0.15 (at 120 Hz, 20oC), ripple current = 2.71A (RMS, 100 Hz, 105oC - EPCOS is a German manufacturer and mains power in Germany is 50 Hz), life = 8000h, dimensions (LS x D x L): 10mm x 30mm x 55mm - https://br.mouser.com/ProductDetail/871-B43547A9477M000
  • C21-22: 2 x Panasonic EEU-FR1A681 (FR series) - 680uF, 10V, 105oC, DF=0.19 (at 120 Hz, 20oC), ripple current = 0.66A (RMS, 120 Hz, 105oC), life = 6000h, dimensions (LS x D x L): 3.5mm x 8mm x 11.5mm - https://br.mouser.com/ProductDetail/667-EEU-FR1A681
  • C23-24: 2 x Panasonic EEU-FR1V181 (FR series) - 180uF, 35V, 105oC, DF=0.12 (at 120 Hz, 20oC), ripple current = 0.71A (RMS, 120 Hz, 105oC), life = 6000h, dimensions (LS x D x L): 3.5mm x 8mm x 11.5mm - https://br.mouser.com/ProductDetail/667-EEU-FR1V181

A small comment on the lead spacing of the chosen replacements for C11-14 is in order. As thebrunx pointed above and also comparing to recap choices on the SRM-T1(S) made by other people, it may be in fact necessary to slightly reduce the lead spacing from 12.5mm to 10mm. The size of the lead holes in the PCB seems to allow for that, and there seems to be no replacement snap-in caps available with 12.5mm lead spacing and up to 30mm diameter, so one has to make do with 10mm LS here.

One may also notice that the diameter of the chosen replacements for C21-24 is slightly larger (6.5mm -> 8mm). Again, this is due to the fact that there seems to be no available replacement caps with at least the same capacitance / voltage rating, the same lead spacing (3.5mm) and 6.5mm diameter - the smallest diameter available within the former constraints seems to be 8mm. On the other hand, there seems to be enough room in the C21-24 positions for 8mm-diameter caps.

Comparing this with an "audio grade" choice of caps:

So, one can see that the Panasonic + EPCOS selection outperforms or at least matches the Nichicon "audio grade" selection on all characteristics, apart from the DF of C9-10, for which the difference seems not to be of much significance (Nichicon KZ: 0.08, Panasonic FR: 0.1). I imagine that the latter is outweighed by advantages of the Panasonic cap in the other characteristics for power supply applications (is it?)... As for the price, yeah, the Panasonic + EPCOS selection is cheaper (notice that the "audio grade" caps linked above are priced in Euro) but not by nearly as much as I thought, so it really boils down to specs and quality for the intended purpose.

One concern I still have (warning: people who didn't bother to read up to here may do so from now on 😁 ) is the following: I understand that increasing capacitances in the recap procedure increases the filtering capabilities of the power supply circuit. However, won't doing this cause other problems if I increase the capacitances too much above stock values? Is it safe to increase capacitance as much as possible just within the allowed physical dimensions of the caps, as recommended in this thread?

  • Like 1
Link to post
Share on other sites

Now, regarding tube replacement / rolling...

I am pleased with the sound of my Gold Aero / Sylvania 6CG7 tubes (they probably can use some rebiasing, though, but that will wait for after recapping), but I have nothing to compare them against, and as I said above it's good to have spare tubes when mine decide to die out, at least. JimL forwarded above advice hirsch and spritzer gave on other threads about Japanese short plate tubes sounding best with the SRM-T1(S) circuit. Digging deeper on these threads revealed that Toshiba 6CG7's seem to be preferred, and that some of these have been rebranded by Raytheon.

  • How do the Raytheon-rebranded Toshibas compare against the "original" Raytheons made in USA in the SRM-T1(S)?
  • How about other Japanese brands, like Matsushita (i.e. Panasonic), Hitachi and NEC (some of 6CG7's made by them were also rebranded by Raytheon, by the way)? How do they fare against the Toshibas on the SRM-T1(S)?
  • Some USA made 6CG7's like the "clear top" RCA (i.e. without the chrome plating on the top of the glass most 6CG7's have) and Amperex (these were made by Philips, I believe) also seem to be highly regarded in general - even more than the Sylvanias by some -, but how do they compare against the Toshibas on the SRM-T1(S)?
  • Does the center shield present in the 6CG7 and absent in most of the 6FQ7's make any sonic difference?
  • Quite a few people recommend going for "black plate" tubes against the "grey plate" ones (like mine). Is there really a sonic difference?
  • Like 1
Link to post
Share on other sites

I've found a potential problem concerning a possible future replacement of the bias trimpots (TVR1-2) in the SRM-T1S. I couldn't find 2kOhm trimpots with at least 0.6W power rating as recommended by JimL above with the same PCB footprint as the original trimpots (three pins forming an equilateral triangle - 3 x 60 degrees). The closest PCB footprint I could find for these is with three pins forming a right isosceles triangle (2 x 45 degrees + 1 x 90 degrees - e.g. Bourns 3345W single-turn). Ideally, were I to replace the bias trimpots, I'd use multiturn parts to make the rebiasing easier as recommended by spritzer and others in other threads, but the only multiturn trimpots I could find with vertical adjustment and the same specs (2kOhm, >=0.6W) have a PCB footprint with the three pins in a single straight line (e.g. Bourns 3252W and 3290W). In view of that, I have the following questions:

  • Is it safe to bend the pins of the trimpots in order to make them fit into the existing PCB holes? I'm unable to see whether the original trimpots have bent pins or not...
  • May I leverage the multiturn option and use higher resistance trimpots so I can have more footprint options? For instance, may I use 3.3kOhm, 1W trimpots instead? These would be compatible with a 10mA current and the rule of thirds. By that logic, these would probably be the largest resistance trimpots I could use, since I couldn't find trimpots with a power rating higher than 1W. Even so, I'm not sure I'd be able to find options with the same PCB footprint...
  • In view of that, how much freedom with the "rule of thirds" do I have with the power rating? Allowing for lower power ratings (1/2W? 1/4W?) would significantly increase my choices...
  • Like 1
Link to post
Share on other sites

Another improvement I've seen people around (e.g. spritzer) suggest on the SRM-T1 is to replace the RCA input jacks and input wiring, but judging from pictures it seems that there is some improvements made by Stax on that area from the T1 to the T1S - the RCA jacks definitely look beefier and the input wiring looks nice enough (actually better than the output wiring)... but again, that's only a visual impression. I wonder whether it's sonically worth the trouble to make these upgrades... Anyhow, I'm using the XLR balanced inputs in my SRM-T1S for now, so this makes upgrading the RCA input jacks moot (but possibly not the input wiring).

I've also seen some people (e.g. spritzer again) recommend to change the tubes from 6CG7/6FQ7 to the modern ECC99 from the Slovakian brand JJ (apparently the only one who makes these), used instead of the 6CG7/6FQ7's in the SRM-600 variant of the SRM-T1 circuit, but that requires rewiring the tube sockets somewhat and changing the R20 resistors in series with the TVR2 bias trimpots from 2.2kOhm to 1kOhm, 1/2W in order to achieve proper bias for the tubes, as JimL suggested here:

https://www.head-fi.org/threads/stax-srm-t1-repair-re-cap-mod.796058/post-14276400

Rewiring the tube sockets for the ECC99's can be done as e.g. in these instructions (in German, sorry):

http://blog.prof-x.de/2019/03/02/stax-vacuum-tube-driver-teil-4-ecc99-tube-swap/

The author combines JimL's CCS mod with the ECC99 mod but he changes the CCS trimpots from 100 to 200 Ohms in order to achieve more easily the required 6mA plate current required to run the ECC99's in SRM-T1(S)'s CCS-modded circuit. The claimed improvement over the CCS mod with stock tubes is an extended frequency response at higher volumes, but given that the ECC99 mod is not easily reversible, should one regret doing it (particularly due to the pair of PCB trail cuts in the rewiring process), I wonder how much one gains with it compared with the CCS mod alone (possibly also after a tube replacement / rolling with some nice Japanese short plate NOS tubes such as Toshiba, Matsushita, Hitachi or NEC). Another thing that worries me about the ECC99 mod is this tube's longevity, which seems to be not so great in certain setups:

https://www.diyaudio.com/forums/tubes-valves/205364-lifespan-jj-ecc99-tubes.html

JimL claimed in the AudioXpress article where he published his CCS mod that SRM-T1(S)'s circuit is rather conservative with the stock tubes, despite Kevin Gilmore's criticism that the 6CG7/6FQ7 tube being driven somewhat above its rated voltage by the SRM-T1 circuit also causes loss of high-frequency response at higher volumes, but how much strain does the SRM-T1 circuit put on the ECC99? More precisely, how is ECC99's lifespan on the SRM-T1(S), particularly with the CCS mod?

Edit (May 12th, 2021): Another tube modification I have seen being suggested (e.g. by spritzer, once more) is to replace the 6CG7/6FQ7 tubes with the USSR/Russia-made 6N6PI (which is said to differ from the 6N6P tube only by a so-called "(im)pulse" operation - no idea what that means) instead of the ECC99. The case for this is that the 6N6P(I) has the same pin layout as the 6CG7/6FQ7's apart from pin 9, which is connected to a screen separating both triode sections in the 6N6P(I) which does not exist in the 6CG7/6FQ7. This screen seems to be disconnected from the remainder of the internals of the tube, which is consistent with spritzer's claim that the 6N6P(I) has the same pin layout as the 6CG7/6FQ7, period, unlike the ECC99 for which pin 9 is connected to a section of the heater. Therefore, tube rolling the SRM-T1S with the 6N6P(I) does not require rewiring the tube sockets as we need to do with the ECC99.

However, according to spritzer in the link above, we still need to replace the 33K/2W plate (anode) resistors with 30K/2W ones since the heater current on the 6N6P(I) is 50% larger (900 mA) than on the 6CG7/6FQ7 (600 mA). Interestingly, it is the same resistor substitution used for the ECC99 (which has a 800 mA heater current)... If one were not to apply the CCS mod as well. Otherwise, one just looses all the plate resistors altogether, which are replaced by the CCS's. If I understand correctly, given that both the 6N6P(I) and the ECC99 require the same plate resistor substitution, the former must use the same CCS current as the latter, as derived from Ohm's law and the 320V rail voltage connected to the anodes (320V / (2 x 33kOhm) = about 4.8-4.9mA per plate goes to 320V / (2 x 30kOhm) = about 5.3mA per plate). Is this correct?

What I fail to understand is which is the possible audio improvement one may get from tube rolling the SRM-T1S with the 6N6P(I) instead of the ECC99, if any. Again, if I correctly undertood Kevin Gilmore's criticism of the SRM-T1(S)'s design, the main problem with using 6CG7/6FQ7 tubes in this circuit is that they end up being driven by a plate voltage too close to its maximum spec (330V, formerly rated 300V), which causes loss of high-frequency response at higher volumes despite the fact that the tube itself is not so strained since it is only being driven at 320V x 9.8mA = 3.1W = 55% of its maximum plate power dissipation (5.7W on both plates together, according to more recent datasheets). Given that the loss of performance caused by this mild voltage overdrive is a lesser effect than the loss of output current in the plate resistor array (solved by the CCS mod), the ECC99 nonetheless solves the former problem since its maximum plate voltage is 400V.

The plate dissipation power with either the ECC99 or the 6N6P(I) is 320V x 10.6mA = 3.4W, whereas the maximum plate dissipation power on the ECC99 is stated in its current datasheet from JJ as 3.5W - per section? It does not say... Older datasheets set it at 5W, some forums reported 8W on both plates but that info is most likely outdated. Update - June 8th, 2021: I've contacted JJ directly to clarify this, and according to them the maximum plate dissipation power of the ECC99 tube is indeed 3.5W per section and 7W on both plates together. That puts the plate dissipation power of the ECC99 in the SRM-T1S at about half of its maximum, which should make for great tube lifespan just like the 6CG7/6FQ7. That kind of answers my second question above. I conclude that the only technical disadvantage of using ECC99's in the SRM-T1S is having to rewire the tube sockets in a not easily reversible (if at all) way, so one is basically stuck with ECC99 tubes if one decides to follow that route. The question of whether the ensuing sonic improvement is worth it remains, of course.

The 6N6P(I), on the other hand, has about the same, slightly insufficient (300V) maximum plate voltage as the 6CG7/6FQ7 and therefore is potentially subject to the same voltage overdrive problem as the 6CG7/6FQ7. Tube longevity should be ok, though, since the 6N6P(I)'s maximum plate dissipation power is 4W in the 6N6PI variant (couldn't figure out either whether this figure is for a single section or both) and 8W for both plates in the 6N6P variant. Why should I expect the 6N6P(I) to perform sonically better in the SRM-T1S circuit than the 6CG7/6FQ7, let alone the ECC99? The only advantages I see in the 6N6P(I) for the CCS-modded circuit are:

  1. Tube rolling with the 6N6P(I) instead of the ECC99 is (at least more easily) reversible (the only other operation needed besides exchanging tubes and rebiasing should be adjusting the plate current in the CCS's trimpots),
  2. Possibly a (slightly) longer lifespan than the ECC99 due to the more conservative plate dissipation power (up to a rated service life of 500 hours for the 6N6PI variant, against the 2000-hour figure of the 6N6P variant - no idea why there is such a difference), and
  3. 6N6P(I) tubes are usually cheaper and easier to find than adequate 6CG7/6FQ7's, especially in matched pairs for easier rebiasing,

none of which really being related to audio quality. Am I missing something?

I suppose much (all?) of the above information has already been discussed at length in the (loooong) Stax threads in this and the Head-Fi forums along several years, but I've found it difficult to gather all of it in an up-to-date form, so I apologize in advance for the redundancies on my behalf in this thread and appreciate the community's patience with them. Again, it seems convenient (at least to me) to have everything I need in one place.

Edited by plaurids
Added new info regarding 6n6pi tubes and ECC99's maximum plate dissipation power
  • Like 1
Link to post
Share on other sites
  • 3 months later...

OK, got all the replacement caps, safety resistors and a nice CCS mod circuit - what I'm lacking now is time to open the energizer and change the components...

Also just got an additional incentive to recap and mod my SRM-T1S - managed to get a fine used Omega SR-007A (serial number SZ2-1636) in addition to my Lambda Signature SR-407. Currently burning in the newly acquired cans, but already amazed at how different the 007A and the 407 sound from each other.

The most obvious differences to me are the noticeably stronger bass and somewhat depressed upper midrange of the 007A as compared to the 407. Resolution, instrument separation and imaging are clearly superior on the 007A (I didn't think that was possible), and the soundstage is wider (probably partly due to the thicker earpads). However, the aforementioned relative "darkness" in the mids and a bit of excessive "blast" in the bass (both to an extent depending on the recording) make the 407 sound more natural to my ears (personally, I'm not really a bass buff), at least when both are paired with my yet unmodded SRM-T1S, but 80% of the time the 007A is actually more pleasant to listen to (the 407 seems to be less forgiving with respect to less than ideal recordings). And yes, I do need a bit more of volume on the 007A - I find myself pushing the volume to 10-11 o'clock on the 007A and to 9-10 o'clock on the 407.

Can't wait to hear how the 007A will sound after the mod on the SRM-T1S.

Link to post
Share on other sites
On 1/21/2021 at 2:36 AM, naamanf said:

Needs Blackgates.

Oh, are you referring to the electrolytic caps? I've already bought the EPCOS / Panasonic combo listed above, I figured it would make most sense since these are all power supply caps and they seemed to outperform or match most other choices for that application. Not to mention that Black Gates are discontinued and hence hard to find in the specs I need. Actually, I did look for Black Gates while researching this, but Hificollective is the only place I could find that still has some of those in stock.

For instance, Black Gate WKz caps would theoretically replace the C11-14 snap-ins but their physical size is too large for them to fit in the PCB, and the ones with the appropriate rated voltage (500V) are only 100 uF at most, so at best one doesn't gain any extra filtering muscle compared to stock. The EPCOS caps, on the other other hand, are 470 uF and still are smaller than stock. Likewise, I couldn't find any Black Gate caps that can match or outperform the specs of the Panasonic caps for the replacements I need within the allowed physical dimensions for the SRM-T1S's PCB.

Honestly, I couldn't get data on DF and ripple current for the Black Gates (which are the other specs that count the most for power supply caps), but I doubt they outperform the EPCOS / Panasonic combo within the allowed PCB dimensions and lower bounds on the capacitances and voltage ratings in that respect either.

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...