New Posts  All Forums:Forum Nav:

SRM252mod - Page 2

post #16 of 21
Thread Starter 

Here are two more measurements of the quiet noise at the amp output with another linear 12v power supply. 

(unmodified (500xgain) SRM252 unit, input poti at minimum, driving SR003). Worse peaks even between 3 and 4.5 kHz.

Shows a relative big influence of the wallwart psu at the amp outputs. 

 

700

 

700

post #17 of 21
Quote:
Originally Posted by 3210 View Post

 

Yes, 22uF/250V.

I measured the silent noise of the SRM252 (with a silent track from CD) driving the SR003.

It shows significant spikes that could come from the HV generation circuit.

 

700

 

700

It's great to know how it looks like. I've sold most of my measurement equipment ~18 years ago, so I little "blind" and only can judge by listening to.

 

Try to replace 470ohm R43 and R42 resistors with a coils/choke having maybe even as much as 1000uH.

 

Try to replace C5 and C4 with better caps (WIMA?) with larger capacity (100n?), solder it at the other side of PCB (bottom), and add some low ESR electrolytic capacitor instead of original (at the top) - like this one http://allegro.pl/kondensator-47uf-250v-12-5x25-low-esr-105-fv-i2921990007.html but in horizontal position, because in vertical fit 10x20mm or 12.5x20mm (most of 47uF/250V have 12.5x25mm and are too tall). After this try to cut path at bottom of PCB between C38, C40 and C4, C5 and put there another two coils/choke (SMD this time).

 

Remove both 330pF C1 capacitors (you may exchange it with 47pF, 100pF) - original 330pF is way too much!

 

I have to remove D2, D3 LEDs (place jumpers instead), and at the same time exchange 100ohm R19, R20 with 560ohm [WRONG value calculated - look at the next post] - it should lower the GAIN in open loop, but I don't know if it helps to tame heavy intermodulation distortion of this amp.

 

A have to do all of this myself to my SRAM-252A, but I still have no much spare time for this project. :(


Edited by borizm - 1/18/13 at 5:25pm
post #18 of 21

I've done most of mentioned changes and I see great improvements to the sound. Very old (year 1979) SRM-1 (improved and modded to Pro) is still at least a little better, but I can enjoy the sound of SRM-252A right now, which was unacceptable before.

 

Input:

Both R1 and both C1 are removed. I have to remove both R2 also and place jumper here (it's stupid thing to buy an expensive cables just to listen to the unnecessary resistors). Hights are clearer.

 

Power:

C31 was replaced by 470uF/16V low ESR. C31 and C36 got also 3u3F/16 X5R (monolotic) SMD.

C4 and C5 are replaced by 2x Jamicon TH (low ESR) 47uF/250V 12.5x25mm (16x25mm is a safe max dimension - I'm looking for low ESR 100uF/250V of this size) supported by 2x SAMWHA X5R 1u/250 SMD 2225 (monolitic) and 2 x Miflex old Polish foil 68nF/250V caps (I'm looking for foil 220nF or 330nF/250V [R10 dimension]).

C37, C39 are replaced by another 2x 1u/250 SMD 2225 (monolitic X5R) SAMWHA.

R42 and R43 are replaced by endure Bourns 470uH/62mA/[15ohm] SMD 1812 coils/chokes.

Between both C4, C5 and C37, C39 both paths was cut and another 2x Bourns 470uH/62mA/[15ohm] SMD 1812 coils/chokes was placed.

This alone was very visible improvement to the "culture"/clearness of the sound.

By the way voltage rose from +209V and -207V to +233V and -230V what is also good.

 

Both channels D2, D3 was removed (jumpers was placed) and path was cut to separate both R19 and R20 from power line by 2Kohm Vishay/Beyschlag MMELF MMA0204 resistors (1x for each channel) - it's perfect value for keep the same currents and no tuning by use of potentiometers is really needed after.

This also has improved the sound (at least I've a feeling of the improvement). By the way, it well known that LEDs are not very linear with small currents ;) , so why Japanese "engineers" made another mistake here, in this amp and in higher amps models?!? Only for the faster visual inspection of the failure in the amp?!?

This modification do almost nothing with the open loop gain value, which I decide to keep as it was.

 

Both D4 ~1.54V RED LED was replaced to ~1.8V GREEN LED, just to rise a bit output current from 1.2mA to something about 1.5mA. This also has improved the sound especially when bass appears with higher volume.

Power consumption (measured before Stax XFF-1240 12V/400mA adapter) has risen from ~6.8W to ~7.5W.

 

R5, R6, R11, R12 have not been touched but I've a plan to replace it to safe for input JFETs values of 5Kohm and 2Mohm - it means GAIN of 400V/V (stock is 500V/V [200mV/100V]), but it may cause side effect, not an improvement.

 

I put the photos in few days.

 

When I finish, I call it SRM-252U (Ultimate) :D

 

P.S.

I think I need second SRM-232A to compare following changes in one of them side-by-side... and I need fast switch box (with two cables ended with 5-pin Stax plugs and two 5-pin Stax inlets) - it will costs me... :D


Edited by borizm - 1/18/13 at 5:28pm
post #19 of 21
Thread Starter 

The quit noise measurements of my modified SRM252 (Tantals+HighCaps+x250gain)

shows slightly more spikes below 3kHz and less above (same for left channel):

 

 

The rectangular response shows no significant difference between the unmodified (252rec_r) and the modified version (252rec2r)

at 160Vrms (I can just measure up to 175Vrms (10:1 reduction), the differences are measurement related):

 

 

 

Quote:
Originally Posted by borizm View Post

I've done most of mentioned changes and I see great improvements to the sound. Very old (year 1979) SRM-1 (improved and modded to Pro) is still at least a little better, but I can enjoy the sound of SRM-252A right now, which was unacceptable before.


When I finish, I call it SRM-252U (Ultimate) :D

 

P.S.

I think I need second SRM-232A to compare following changes in one of them side-by-side... and I need fast switch box (with two cables ended with 5-pin Stax plugs and two 5-pin Stax inlets) - it will costs me... :D

 

A friend of mine says my modified version sound more clear and precise than the unmodified one.

He also prefers it in comparison with the actual SRM252 (x800gain) version.

He may try the gain reduction with the SRM323 (the actual x800 version has two 150kOhm resistors in series,

so it's easy to test it (at -6dB max) for the bigger Stax earspeaker by bridging one of them).

 

I finally changed the 22uF/250V caps with 22uF/350V (same size but lower tan(phi)),

mounted plastic feet below my 252 (8mm high, improves heat dissipartion)

and engraved "SR-003 ONLY" over "Pro Only" above the headphone jack 

so it's still pretty original but also special.

post #20 of 21
Thread Starter 

The switching frequency of the HV generation circuit seems to be 72 kHz

(with C33=C_T=10nF and R33=R_T=1.8kOhm, see SG3524 datasheet).

The bandwidth of the amplifier should end way below 72kHz,

but reducing the gain will increase the power bandwidth,

so I don't recommend my feetback mod for the SRM252.

post #21 of 21
Thread Starter 

Hi Mark!

This converted SRS005 set is now in operation. Punchy bass, decent mids and clean highs excellent.
The notch filter in the CS4344 DAC (3.9 mH-SMCC + 180pF mica) reduces 4fs peak (176kHz) to remarkable 12dB
and reduced overall even the noise spectrum of the Jensen transformer (Ci-2RR 'Isomax') in unexpected ways.
Really nice, a 20kHz sine measures but only after Jensen (with its 50kHz bandwidth).
The maximum input signal level is reduced by Jensen of 0.65 Vrms to 0.54 Vrms,
i.e. on computational 135Vrms (x250gain) at the output of SRM252mod (corresponds to-6dB with respect to 280Vrms).
In normal hearing, the potentiometer is at 13.30 clock (with decent headroom).
The max. Overshoot of the CS4344 is about 50%, so that the enough yet
Power reserve (as well as safety margin) is.
The bass (punch-) - and height loss (<1 dB at 20 kHz) by the Jensen transformer
is the input resistance of the SRM252 25kOhm m.E. negligible.
Only in this combination (with modified CN Optrca10 DAC and Jensen CI-2RR)
I think for sure the SR003 on SRM252mod operable.

Best regards,

(translated via Google)


Hallo Mark!

Das umgebaute SRS005-Set ist jetzt in Betrieb. Punchiger Bass, dezente Mitten und exzellent saubere Höhen.
Der Notch-Filter im CS4344-DAC (3,9mH-SMCC + 180pF-Glimmer) reduziert den 4fs-Peak (176kHz) um bemerkenswerte 12dB
und verringert insgesamt sogar das Störspektrum des Jensen-Übertragers (Ci-2RR 'IsoMax') in unerwarteter Weise.
Richtig schön misst sich ein 20kHz-Sinus aber erst nach dem Jensen (mit dessen 50kHz Bandbreite).
Der maximale Eingangs-Signalpegel verringert sich durch den Jensen von 0,65Vrms auf 0,54Vrms,
d.h. auf rechnerische 135Vrms (x250gain) am Ausgang des SRM252mod (entspricht -6dB bzgl. 280Vrms).
Beim Normalhören steht das Poti auf 13.30 Uhr (mit ordentlich Pegelreserve).
Das max. Überschwingen des CS4344 beträgt ca. 50%, sodaß die noch genug
Spannungsreserve (sowie Sicherheitsreserve) besteht.
Der Baß(-punch)- und Höhenverlust (<1dB bei 20kHz) durch den Jensen-Übertrager
ist am 25kOhm Eingangswiderstand des SRM252 m.E. vernachlässigbar gering.
Einzig in dieser Kombination (mit modifiziertem KN-Optrca10-DAC und Jensen-CI-2RR)
halte ich den SR003 für sicher am SRM252mod betreibbar.

Viele Grüße,

New Posts  All Forums:Forum Nav:
  Return Home