Modifying Stax's SRM-T1

[spritzer]

Hangikjöt is delicious but I don't touch any of the other shit.  My GF does like svið which is kinda revolting... 

[Craig Sawyers]

Not really relevant to the thread, but kinda related to old circuit boards. I was recently fixing a friend of a friend's Michaelson Odysseus power amp - which had a single sided board. Apart from a new set of EL34's I noticed a power resistor that would wobble if poked it. Turned the board upside down, and this is the horror story I found. Lots of thermally marginal parts too close to the board, resulting in toasted board, and multiple levels of repair by others. Needed two 47 ohm power resistors one of which was O/C, and both of which were 2.5W parts dissipating 3W. Replaced them with 7W jobbies.

Edited September 11, 2017 by Craig Sawyers

[kevin gilmore]

Mikhail called that problem "Acetate Failure"  Every singlepower extreme, and many of the others had diode bridge and pass transistor burning

Edited September 11, 2017 by kevin gilmore

[spritzer]

Ahh yes, the old acetate failure... must have been the same problem as with the 12.6V transformer used in the Extreme's...   

[Craig Sawyers]

Acetate failure - must remember that 

The casework also runs stinking hot. The four EL34's for each channel are horizontal between a slotted bottom panel and a perforated cage. I thermocoupled the cage - 80C (175F) and easy enough to be hazardous.

This particular amp was only made for two years because the failure rate was so high that many did not even get out of the factory without needing their first repair. The only reason the guy had this amp was Anthony Michaelson gave him it (he'd done that industrial appearance design for Michaelson; although the electronic design is marginal to say the least, it does look good)

Edited September 11, 2017 by Craig Sawyers

[kevin gilmore]

1 hour ago, spritzer said:

Ahh yes, the old acetate failure... must have been the same problem as with the 12.6V transformer used in the Extreme's...   

that was actually the enamel failure. completely different thing.

And those were 6.3V transformers with one of the primaries wired as a secondary

 

Edited September 11, 2017 by kevin gilmore

[spritzer]

Must try that sometime.  Use a transformer way past its rating but only use one winding...  Yup quality stuff. 

[JimL]

Speaking of old circuit boards, I remember doing mods on Dynaco phenolic PAS-3 boards.

[Mach3]

On 9/10/2017 at 9:38 AM, spritzer said:

Having a T1S sitting idle here I decided to have some fun with it.  Drew up a small CCS board, a scrap heatsink and voila...an improved amp is born.  Bias supply fixed, all caps replaced and output resistors fitted.  I ran it a bit hotter than stock so I had to modify the tail resistor on the output cathodes down to 1.5K but that is easy. 

Trying to do this mod this my SRM-T1S and SRM-600. By tail resistor, did you mean R20 stock 2.2k to 1.5k?

Also I'm looking at your custom CCS PCB and comparing it to Kevin's one. Where are the four DN2540 and there 2 giant WIMA cap.

Or are they hidden behind the large heat sink.

[Something Incongruous]

@JimLFor starters, thank you very much for sharing. Have since found your article as well, so i'd consider me in a good place to start. And given my non-existing skillset, this is where you may all laugh heartily

Am a recent Stax convert in search of a amp. Have had a lot to weigh in and eventually had it down to one out of three; i) a used T1 with your mod, ii) a used 727 -unmodded for starters- (no offense to the usual two suspects), iii) a used 353X/XBK with only a heatsink replacement, largest i could fit. Given availability, pricing and most of all condition-to-price, i 'settled' on buying a used T1 but.. i'm no longer sure buying a battered 5-hands down piece of equipment that's seen, what? Thirty years of usage? for 600ish is wise. Given my limited budget that is (it's not only the initial cost, it's also the longevity i need take in consideration).

Since the newer T1 version diagram is also available -is that even legal? Tell me not-, i've decided i will start from scratch; draw the PCB with your modifications incorprated, print it and slowly go from there. Which really adds to it all in some ways, doesn't it ^^

Opinions/ideas as to the.. soundness of my (final, revision 2.0) approach are more than welcome.

 

* And speaking of ideas,

On 6/28/2017 at 5:51 PM, kevin gilmore said:

work in progress

hangs off the side with thin angle bracket, no drilling of holes

 

( @kevin gilmoresince i'm tagging you anyway, not sure if i've ever thanked you before, so, sincere thanks for making so much of your work available. Leaves options for folks like me that we wouldn't ever have had otherwise. Much obliged sir)

Was wondering if the above's progressed any further? And if so, would you be willing to share?

[canyoncruz]

Concerning the CCS,' I fond this else where and does this pertain to the low current draw in this circuit?

Figure 2.62 COSS output capacitance of the DN2540.
As can be seen from the graph, COSS is alarmingly high at low voltages and it is
not until Vds>15 V that it falls to its asymptotic value of ≈12 pF. Fortunately,
we would always operate at Vds>15 V (preferably 40 V, or more) because a
BJT cascode is a better choice at lower voltages. An EF184 pentode CCS could
achieve an output capacitance of 3 pF, but once we add typical strays of 3 pF to
both circuits, the valve advantage degrades to 6 pF against 15 pF. This
somewhat poorer output capacitance is the price we pay for the undoubted
convenience of a two-terminal CCS.
The device has a maximum continuous current rating of 500 mA, so it should
come as no surprise to learn that performance degrades at low currents. If you
need a current <10 mA, then you owe it to yourself to see if there’s an
alternative solution because the device is much better >10 mA, and at 25 mA it
really comes alive ( rout of a cascode CCS at 25 mA was four times that at 10
mA, all other factors kept constant).
References

 

[JimL]

Sorry for the delay in replying. Your quote is from Morgan Jones excellent book on valve amplifiers. He is talking about cascoded CCS using DN2540 MOSFETs as both the upper and lower devices in a CCS. Here the issue is, as he notes, that the lower device sits between the gate and source of the upper device, and sees a low voltage, and the COSS output capacitance is relatively high until the voltage across the CCS reaches about 15V.

 

Part of the reason for this is that when you use a DN2540 for the upper device, the GS voltage is quite variable, between 1.5 and 3.5 volts. In this case, higher is better, since a higher GS voltage puts the lower device into the saturation region, where the current changes very little with variations in voltage, and the device acts as a current source. Using a 10M90S as the upper device is one way to achieve this, since by design, the GS voltage of this MOSFET is at or above 3 volts for currents between 1 and 20 mA, whereas with a DN2540 as the upper device, the lower DN2540 could see a GS voltage as low as 1.5 volts which is much closer to, or below the point where it drops out of the saturation region. 

However, more significantly, since we are using the cascode CCS in the output stage, the standing voltage across it is about 320 volts. Even if we assume we are using the DN2540 for both devices (a bad idea since the total voltage across it could exceed its 400 volt maximum resulting in instant annihilation of the CCS), the only time the DS voltage would decrease to less than 15 volts, resulting in a higher output capacitance, is when the output voltage for half of the output stage reaches a peak of (320-15) x 2 (or supply voltage - 15, doubled because the other half of the output stage is at the opposite extreme voltage) = 610 volts peak = 430 volts RMS. With the typical Stax headphone such as an SR-007 Mk II with a rated sensitivity of 100dB SPL/100 VRMS/1 kHz, this would produce an SPL of 127 dB, close to the 130 dB threshold of pain.

 

So basically, because the CCS sits at a high voltage all the time, it never sees the conditions that Morgan Jones is concerned about except when SPLs are so high that you risk permanent hearing damage. I wouldn't worry about it.

 

See also my following comments below.

Edited December 11, 2020 by JimL

[JimL]

A simple way to look at the Morgan Jones COSS graph is that the cascoded CCS requires a minimum voltage to operate optimally - in the case where both devices are DN2540 MOSFETs, at least 15 volts. As long as that criteria is met, no problem.

 

Also, even at lower currents, the cascode performance of the 10M90S/DN2540 is excellent because the upper 10M90S device maintains the DS voltage for the lower DN2540 at around 3V even at 1 mA. That device is specifically designed as a current source, to keep a very constant GS voltage for currents between 1 and 100 mA regardless of gyrations in DS voltage. Remember that the Morgan Jones comments refer specifically to the use of the DN2540 as the upper as well as the lower device.

Edited December 11, 2020 by JimL