Melos SHA-Gold "Maestrobator"

[spritzer]

One of these rare amps stopped by for a visit:

This one belongs to Stuart but I'm taking a look at it for him to sort out the unbelievably noisy CT2 stepper and do something about the very high gain. I was thinking about a simple voltage divider on the input instead of messing with the circuit. I might also change out the PSU caps as some do appear to be stock. Any further ideas about this beast? I did read Doug's thread on the SHA-1 but I doubt I'll do something as extreme. It very noisy though and quite frankly, doesn't sound very good with RS-2's and Mullard E88CC tubes. Very edgy and with crappy bass. There also appears to be a small channel imbalance.

[nikongod]

The stepper was dirty when I saw that one. The channel balance should return to normal when the stepper is cleaned.

I used a pre-attenuator on my melos. I forget how much gain they have stock. I think its voltage gain=7 aka a lot

link to pre attenuator

[kevin gilmore]

there are so many design flaws in that thing i don't know where to start.

but the biggest one is using the filaments as the current source for the output stage.

Once you hack that out, and run the tubes the normal way, and use a real resistor

for the current source, lots of problems go away.

[spritzer]

Stuart did say you looked this one over. Thanks for the link, I was doing this the hard way by inserting a pot at the input, adjusting it for the range I liked and then measuring it.

I'm not really looking into any major mods for it but does anybody have the schematic? Using the filaments is just a terrible idea...

[Salt Peanuts]

For all the positive posts/threads Melos received over on HF, I remember being thoroughly underwhelmed when I finally got to hear one at a meet.

[kevin gilmore]

I'm not really looking into any major mods for it but does anybody have the schematic? Using the filaments is just a terrible idea...

silly boy...

http://gilmore.chem.northwestern.edu/melos.jpg

[dsavitsk]

As far as gain goes, you are probably better off replacing the 6dj8 with a lower mu tube if that's an issue. It does not need to drive anything very difficult, so increasing the rp is fine. Biggest issue in doing this is the heater which may mean some creative trace cutting.

or all the positive posts/threads Melos received over on HF, I remember being thoroughly underwhelmed when I finally got to hear one at a meet.

Compared to what was available at the time, it was a nice amp. But, there are lots of much better amps out there today.

[nattonrice]

That pcb is such an asshole to work on it isn't funny.

Birgir, apart from all the permutations of different capacitor and resistor substitutions/upgrades the biggest audible change came from the left channel mod.

The one I did went from being a noisy piece of shit to quite listenable.

If it isn't done already I'd definitely do it and leave it at that... possibly change out the HV bridge rec for something decent.

[spritzer]

silly boy...

http://gilmore.chem.northwestern.edu/melos.jpg

Sorry about that... The filament wiring is just too funny, with traces running all over the board.

As far as gain goes, you are probably better off replacing the 6dj8 with a lower mu tube if that's an issue. It does not need to drive anything very difficult, so increasing the rp is fine. Biggest issue in doing this is the heater which may mean some creative trace cutting.

That should also solve some of the noise issues. The plan is to remove the filaments from the main circuit so that is a non issue really.

That pcb is such an asshole to work on it isn't funny.

Birgir, apart from all the permutations of different capacitor and resistor substitutions/upgrades the biggest audible change came from the left channel mod.

The one I did went from being a noisy piece of shit to quite listenable.

If it isn't done already I'd definitely do it and leave it at that... possibly change out the HV bridge rec for something decent.

I think the mods here were just better parts and some very expensive tubes (125$ each) with no care taken to just how messed up the PCB is.

[spritzer]

While weighing my options (and asking Stuart how far he wants me to take this amp) I figured I should just fix what was really broken. I removed the DACT stepper and stripped it down completely, polished the metal contacts with Auto-sol and then bathed them in De-oxit and Pro-Gold. While I was at it I added a voltage divider directly to the CT2.

I didn't have the exact values but it is close to 10dB. Now that it is back in the amp the stepped is working 100% again and the lower volume level is a welcome change.

[nikongod]

A 12au7 will fit with 1 trace cut & 2 splices to run the heaters differently. The original prototypes of the Melos used a 12au7 as both a simple lame common cathode (lame) and as a LTP. I am pretty sure that the 6dj8 runs better with ~40V & 0.75mA across each section than the 12au7 does... I would not try it myself.

Most of the noise on the sha-gold comes from the B+ supply which crosses the PCB straight through the middle of the audio supply twice. I suspect there is a fair bit injected by the op amp DC servo, but have not tested this.

The heater supplies are regulated DC, Ears I trust say filaments should be heated by CCS supplies or full level AC. Nominal levels of signal on a heater separate from the cathode isn't going to hurt anyone, people use full AC with no hum! Most weak attempts at DC heaters with a single big cap after a rectifier have more noise than the signal used to run a headphone. and its nasty spiky noise too.

Im not sure if Stuart's Melos was moded before or after the left-channel mod was "created" It didnt sound like it needed it when I heard it.

[spritzer]

Thanks for that Ari. The ECC82 was the alternative I was looking into but since I didn't have any at hand I dismissed the idea.

This one has an access panel in the bottom so when I turn it off now I'll take a look if there are any wires underneath the PCB. It's a bit odd though that there are two wires going to the PCB from the pot, per channel. Both go underneath the board so I don't know where they go exactly.

I for one have had no issues with AC heaters in the past so that is an option. Not sure what voltage the transformer outputs as this might be a custom model. Not a single marking anywhere except for numbered terminals. One odd thing, the channels were reversed inside the amp. Might have been part of the original tweaks...

[nikongod]

This one has an access panel in the bottom so when I turn it off now I'll take a look if there are any wires underneath the PCB. It's a bit odd though that there are two wires going to the PCB from the pot, per channel. Both go underneath the board so I don't know where they go exactly.

I for one have had no issues with AC heaters in the past so that is an option. Not sure what voltage the transformer outputs as this might be a custom model. Not a single marking anywhere except for numbered terminals. One odd thing, the channels were reversed inside the amp. Might have been part of the original tweaks...

There are lots of wires under the PCB. its icky. All the nastyness on the top of the PCB is pretty much the same on the bottom, plus the wires for stuff they forgot to put in. The left channel mod is a wire from one of the jfets between the tubes to one of the mosfets. The trace near it (the trace is kind of S-shaped, and wraps around the power supply wires ) should be cut at both ends.

The mystery wires may go to one of the rear preamp outputs. The original SHA-gold used the photentiometer as a buffered passive preamp, and sent that signal to both the tubes, and the outputs in the back.

The stock transformers were all over the map I think Stuarts melos does have one of carlos's realllllly nice custom would transformers though. IIRC they had white tape around them rather than the common clear plastic tape.

Ideally the transformer should have:

~120v @100ma for the B+ supply

2*9-12v@~1a for the LV supplies.

[spritzer]

I just spoke to Stuart and he gave me the green light to do what ever I wanted to. This amp does indeed have the custom transformer with white tape around it. Having no markings at all made me more than a little worried when rewiring for 230V...

So the plan now is rewiring the tubes for AC and add a small transformer to power them, install a resistor instead of the filaments, change all the rectifier diodes, replace the rest of the caps and try to separate signal and power the best I can by cutting traces. Perhaps a choke in the PSU would be a good idea as well?

[nikongod]

I would keep the tube heaters/mosfet source loads the way they are now.

They are heaters, not filaments. Within reason you can do whatever you want to a heater, nothing happens. The tiny bit of signal running on the heater does nothing.

If you want to cure everything in the Melos easily, I would modify a Cavalli EHHA to be more like a Melos.

So the plan now is rewiring the tubes for AC and add a small transformer to power them, install a resistor instead of the filaments, change all the rectifier diodes, replace the rest of the caps and try to separate signal and power the best I can by cutting traces. Perhaps a choke in the PSU would be a good idea as well?

Good stuff for upgrading the diodes and caps. Separating signal and power is going to be a BIG task: Aside from the left channel mod its not *that* bad.

I tried putting a choke in the PSU. It caused my amp to randomly oscillate so I took it out. I think Doug tried it too and had the same poor results.

[Knuckledragger]

What is the difference between a heater and a filament in a vacuum tube?

[nikongod]

What is the difference between a heater and a filament in a vacuum tube?

They serve sort of the same function in a different way.

Sorry to be the language-police on it.

Filaments are in directly heated tubes. 300B, 45, 307a etc

Heaters are in Indirectly heated tubes. 12au7, 6sn7, 6dj8, etc

The cathode of the tube must be heated for the electrons for flow through the tube.

A filament is BOTH the heating element, and the cathode in 1 part. this allows for greater heating efficiency than the heater which was very important when everyone used batteries to run their junk. All of the power to heat AND the audio signal goes through the filament. The problem is that the audio signal mixes with whatever noise is on the filament power supply. Everyone loves DHT's though, so we come up with interesting ways to get around the problems.

A heater is electrically insulated from the cathode. This is not as efficient as the filament arrangement but the heater does its thing by its self, the cathode does its thing by its self. There is a TINY bit of interaction between the 2, but by and large they are different parts so noise pickup between the 2 is GREATLY reduced. Separate heater/cathode also allows the heaters and cathodes to be at different voltages which makes pretty much everything easier and less expensive.

Edited November 23, 2010 by nikongod

[Knuckledragger]

Thanks, that clears it up. Language policing is not a bad thing in the technical world.

[spritzer]

Indeed they are heaters. It's the same word in Icelandic for both of these so I do mix it up all the time.

The plan is to get rid of the hiss which so bloody annoying. Now I'm never going to reach complete silence but making it mostly inaudible would be a good start.

[nikongod]

Try switching the DC servo op amp.

Its the only thing I haven't tried.

I guess I should try it too

Be sure to check the voltage offsets, and zero them if necessary with the trimpots near the op amp.

[dsavitsk]

I for one have had no issues with AC heaters in the past so that is an option.

If you do this, make sure you can put it back. I think it will be noisy.

Perhaps a choke in the PSU would be a good idea as well?

I tried putting a choke in the PSU. It caused my amp to randomly oscillate so I took it out. I think Doug tried it too and had the same poor results.

Yes. It was the interaction of the choke + the regulator that sounded like a police siren. When I put in a new regulator (LR8 + a pass transistor), all was well. And with mine (a SHA-1 so maybe different), basically all noise was from B+ and the tubes. The tubes I just had to dig around to find a reasonably quiet pair, but replacing that horrid B+ regulator they used made a big difference.

[nattonrice]

I too discovered I can't randomly slap chokes where ever I want.

Mind you the oscillation set the LEDs doing some pretty cool things.

[spritzer]

On my todo list is designing a new PSu for amps like the GES so doing something from scratch is always an option.

If you do this, make sure you can put it back. I think it will be noisy.

Could very well be but isn't it the modder tradition to receive something that works just fine and send back something that doesn't? I could always do DC filaments Rudistor style, 7808 regulator with a resistor to drop the voltage...