The Multi Amp aka Dynalo Mk2

[Pars]

Helps as I can mark that up. I was looking for something that had Q1, etc. (the part IDs) labeled.

[luvdunhill]

Helps as I can mark that up. I was looking for something that had Q1, etc. (the part IDs) labeled.

That's not the mafia way

[kevin gilmore]

each ssdynalobal board pulls about 180ma at 20v

[Pars]

Marking the jpg up that GrindingThud posted, I arrived at this.

 

It appears that the board has Q1/Q3 and Q2/Q4 of the THAT340 swapped compared to the schematic. The + amp section is on the left side of the board, - amp section on the right.

 

[kevin gilmore]

the schematic capture auto numbers stuff, and I never pay any attention to that when I do the board.

 

what I should do is use the netlist, but that is just so much work.

 

the problem happens when you go back and change the schematic, and the board no longer

matches...

 

if I actually did this for a living, I might pay more attention.

 

then again, for things like the step attenuator where a schematic never existed in the first place...

[spritzer]

Yeah, schematics are for the weak.   

[Pars]

^ Uhhh, yeah

 

At any rate, just reporting what I found. It makes perfect sense from a board layout perspective to have swapped the devices used for each of the complementary pairs as they are on the proper side of the chip to be the closest to the output side they are handling.

 

BTW, what layout software are you kids using?

[luvdunhill]

the problem happens when you go back and change the schematic, and the board no longer matches

Seems like a flaw in the software. Very simple to do this using Diptrace.

[Pars]

A question on paralleled output sections such as that used in the multi-amp:

 

If the gains of each of the 4 complementary pairs are matched closely, as well as between the pairs, is the goal to have equal currents thru each of the 8 legs (4 + and 4-), or equal voltage drops across the emitter resistors? I know that if everything was perfectly matched, those would be one and the same. Of course, nothing is perfect

 

I guess in a nutshell what I am asking is that if I notice one pair having a higher drop across the emitter resistors, would I want to slightly increase the resistance to achieve the same current, or lower the resistance to get the same voltage drop?

 

Anal question, I know

[Pars]

^ No comments on the above, eh?

 

Almost had an oopsies testing board #1. Fired it up while watching offset (+ and - to ground). Hmmm, 1.x volts, doesn't look good. Caught a trace of smoke and shut it down. After verifying that it was the only sets of emitter resistors I had put in (1 pair per side) giving off a bit of smoke, I was looking things over and realized I had forgotten to put the 255 ohm R38/R39 in, and was running 7.x volts across the emitter resistors. After putting 255 ohm in for starters, things looked much better. No adjustment, there is ~40mV offset +/- to ground.

 

I think I'll replace the emitter resistors that are present anyhow. Currently only dropping about 200mV across the 20 ohm emitter resistors, so only 10mA. I'll play around with R38/R39.

 

I had bought a bunch of the Xicon 271 series 0.25W resistors for the emitter resistors so I could match them. I also have some Dale CCF-55 which are 0.25 / 0.5 W rated that I had forgotten about. Any preference as to which YOU would use?

Edited March 20, 2015 by Pars

[Luke]

I know its a little off topic, but how do I know if my HP can be modded for balanced? 

[Mister X]

Cut the plug off.

If there are 4 wires a new plug is all that is required.

You will have to rewire the phones If there are only 3 wires.

[Pars]

If you have 4 wires, ohm out the tip of the plug. That wire is L+. RIng would be R+. Then measure that color wire (L+ and R+) on the portion of the cable going to the phone to the two ground wires. The one that shows some conductivity is the - for that channel (L- or R-).

 

As Mister X mentioned, if you only have 3 wires, time to rewire the phones. I like the Mogami 2893 mini-quad myself.

 

If you need an adapter cable to use the phones with a unbalanced amp, keep the old plug and the chunk of cable and put a 4-pin female inline XLR on it.

Edited March 21, 2015 by Pars

[Pars]

Testing board #1 and all looks good. I think for the time being I will use 301R for R38/R39. This is giving me around 360mV drop across the emitter resistors, so 18mA of current. Others I have tried:

 

no resistor: not recommended

 

370 ohm: 520mV across the emitter resistors, so around 26mA

316 ohm: 410mV across emitter resistors, 20.5mA

 

I may end up using the 316 ohm, but it is still kind of cold in my basement and I would want to see how these did cased up.

 

Still not sure about the pots. RV3 seems to do little or nothing. RV1 and RV2 seem to need to be balanced for best offset + or - to gnd. Once these are dialed in, the + to - offset seems to more or less take care of itself. With no servos in, the + or - to gnd offset seems to float around about 2-3 mV.

 

These servos seem to be more aggressive than what I recall from the original dynalo. Using TL071 opamps, offset is held below 1mV, and gets there quickly, even in earlier testing before I had things adjusted better.

[kevin gilmore]

rv3 is the common mode offset pot. with the 2 other pots in the center, and the servo's

removed you adjust for minimum between both outputs and ground.

then you adjust the other 2, then add the servos

[Pars]

I played around with RV3 today, and it worked as described. The other 2 pots should be close to center, so I left them alone. These seem to take small adjustments, whereas RV3 takes a considerable amount of adjustment, which is probably why I didn't think it was doing anything previously.

 

At first I messed around using RV3 to try to provide equal base voltage on Q6 and Q9, measured at R21/R25. Offset looked terrible.

 

Then I tried the same thing, only with the collectors of Q6 / Q9 (measured at R1/R2 and R23/R4... whats up with that numbering anyhow ). Offset again looked terrible.

 

I centered RV3, then monitored offset on both the + and - to gnd, and started turning RV3. It got the adjustment quite close; close enough that I didn't bother with the other 2 pots. Without servo, it was drifting ~4 or 5 mV total, most of the time more like 2-3mV. With servos, easily within +/- 1mV.

 

I let it run for about an hour. At 365mV 18mA, the board barely gets warm. I even put a box over it to see if it would heat up more, and it didn't. I still think I could crank the current up higher, at least to the 25mA range. At that point I would probably replace the output emitter resistors with something bigger than the CCF55s I have in now.

Edited March 23, 2015 by Pars

[Pars]

Probable stupid question time:

 

I was playing around with board #1 and ran a test signal into it using a Discman (SE, I- tied to gnd). On the outputs, I was seeing 2 signals (O+ and O-) that were each around 2.5Vpp, 180 degrees out of phase. The input signal was from the line out. With a gain of 6 in SE mode, I would have expected the output signals to be larger? Or do I need a load resistor across the outputs (O+ to O-)? Or is this potential loading of the input impedance?

 

Gain resistors are the stock 5K and 25K. This is the first time for me playing around with balanced signals also.

Edited March 24, 2015 by Pars

[kevin gilmore]

likely ,5 volt in which makes sense, gain of 5

[gepardcv]

I'm finally at the point of wiring the amp together, and want to clarify the correct way to handle grounding from the volume pot (I'm using the TKD 601 board to mount the volume pot). Do I run ground wires from the volume pot to the star ground (four in all), or from the volume pot to the amp boards, along with the + and - signal wires (and ignore the ground inputs on the amp boards)?

[kevin gilmore]

i wire 3 wires from each xlr jack to the pot for each channel, then the pot to the boards

not a star ground, but works, and is completely silent

the wires from the xlr jack are shielded triax

[gepardcv]

Thanks! Got it.

 

Next question: I brought up the linear power supply (a Bel Power Linear HAA15-0.8-AG), and I can hear a faint buzz from the transformer when it's on. I doubt it'll be audible once music plays, but still: is there anything I can do to reduce the noise?

[GrindingThud]

I unscrewed and wiggled the xformer then tightened back up. Took care of most of it.

Thanks! Got it.

 

Next question: I brought up the linear power supply (a Bel Power Linear HAA15-0.8-AG), and I can hear a faint buzz from the transformer when it's on. I doubt it'll be audible once music plays, but still: is there anything I can do to reduce the noise?

[gepardcv]

All right, I finished my first Dynalo Mk2 build. Photos: http://imgur.com/a/GSxFt.

 

Have now listened to a few tracks, so far so good with the HD600, completely silent. However, when I connect a sensitive headphone, like a JH13 to it (which yes, I know, doesn't need this kind of amplification), it has a pretty loud hum. Grounding problem somewhere? Could it be isolated to the SE output? I don't have balanced cables for any sensitive equipment I have. I grounded the TRS sleeve to the chassis, is that wrong?

[kevin gilmore]

what I do is float everything in the chassis. The chassis is connected to earth ground.

everything else floats and the audio ground does not touch the chassis ground.

this is easy with xlr connectors as the inputs only, and then rca to xlr adapters for

single ended.

[gepardcv]

A few tweaks later, I made it better. Still not dead-silent with the JH13fp, but that might be too much to ask (28 Ohm, 116dB@1mW), but much less obnoxious than before. I wonder if the headphones pick up noise from the transformer? I tightened the screws on it as GrindingThud suggested, but it didn't help much. That's a little disappointing. I'll follow Kevin's advice about isolating the audio ground from the chassis ground next time I build something like this.

 

I'd like to emphasize that the HD600 sounds great from both 4-pin XLR and SE outputs (I don't have a dual 3-pin XLR cable handy to test right now). Perhaps it just doesn't have the sensitivity to pick up the noise. The amplifier itself rocks. I'm listening to Janos Straker playing Bach's cello suites right now, and loving it.

 

Many, many thanks to Kevin for making and sharing this design. Also thank you to everyone who answered my questions while I put this together.