KG Balanced Dynahi build discussion thread

[rumina]

yes the orientation is fine of the opa445 and i also tested 2 seperate opa445 i had at hand. what puzzels me is that the vf of the one led drops 0.1v when the opa is in, is this a part of the servo function? after my first troubels i changed both leds to the LTL-307ELC, the vf is a tad lower then the clear led i used but easy to correct to the target bias changeing the 500 ohm resistor (now use a 357 ohm for bias around .575v). both led types dim with the opa in (on both boards).  i also changed the that340 to 2 others, the result is the same with the servos. but im happy, the amps is running at the moment an with the lid closed the offset on both channels are stable at under 2mv (with opa the offset multipies by factor 20). god that i machted 80 mje's and 100 pzta's, the amps sounds perfect and is noise free. maybe someone with the same board in the future has a better understanding than i have .

[oal100]

if I understand the schematic correctly, the opa445 acts as an integrator. Could it be that if the parts are well matched - and the opa445 is not in - there is a steady decline of the offset? Which in turn would mean, that the opa would correct all the time in the same direction with a declining rate? And thus overshoot?

Therefore the whole system would start to swing slowly but instigates over time even to V+?

How would one have to measure the amp with an open loop to get at the bode diagram? Would a bode diagram help to find out if the system - in that particular constellation - is getting unstable? Because maybe someone could post it?

[Pars]

If I understand an integrator correctly, and opamps, the opamp is always trying to make its inputs equal. Since the offset is presented on the neg input of the opamp, it will be correcting in the opposite direction to the offset. When I do offset measurements on one of these amps, I've never seen the servo try to go the wrong way...


Sent from my iPhone using Tapatalk

[oal100]

yes yoúre right. but it may oscillate depending on the config. did you see the offset always dampening over a longer period (e.g. 60mins - with no input signal) and being stable or the opposite? the freq around a zero-offset may well be 1/300 Hz or smaller so its hard to see that the amplitude is in fact getting bigger? it sounds to me like getting sea-sick but slowly. therefore i wonder how to measure and falsify this hypothesis...

[Helium]

I think he needs to try another opamp instead of opa445. Maybe DC must be decreased temporarily for another lower voltage opamp, but worth to try.

[penmarker]

I am planning to build using the most recent board and I've been in contact with Dr. Gilmore and several other users here. Luckily a friend of mine is sending out some PCBs for etching so I can tag along with him.

Annoyingly enough, case size needs to be very big to accommodate both amp and PSU, with heatsinks. The cost really starts to hike up from thereon.

[jose]

Yesterday I was listening to my amp and I noticed a small sound coming from both channels, it's not constant and it's not perceived while listening music. It's a sound like "fried".

I was listening SE input and SE output and I don't bridge the cold input to ground.

I have been taking some measurements and BIAS is symmetric on both sides; 750/760 mVolt on R channel and 720/735 mVolt on L channel. On the contrary, offset drift a lot and I've never been able to keep it on 0 volt. It's always between -100~40 mVolt and 120~60 mVolt (without servo). With servo, the offset never exceed 20 mVolt.

It has always seemed to me that Mosfets run too hot for an on-board assembly (Mosfet reach 90°) although I know that the maximum temperature is 150° but even so, components and the PCB also get very hot.

I'm not sure where to start since I've changed the THAT and OPAs too and I still have the same noise.

Can Mosfet be damaged? Or can the problem be elsewhere?

Thanks

[spritzer]

Why would you not ground the - input?  It should never be left floating like that. 

Ohhh and what mosfets are you referring to?  The Dynahi should have bipolar outputs...

[penmarker]

Speaking of grounding, should the each ground and -ve input be grounded to their respective boards or it doesn't matter? Since SE has both grounds tied that means both boards will have their -ve and ground tied together.

[jose]

Why would you not ground the - input?  It should never be left floating like that. 
Ohhh and what mosfets are you referring to?  The Dynahi should have bipolar outputs...

Because at the moment I haven't had problems with that. If it's the problem it's easy to solve but I doubt it's my problem.

As for the output device (hope you like it more...), you're right I'm wrong they aren't mosfet.

[spritzer]

So if the - input is floating, you will get a whole host of issues. 

[jose]

So if the - input is floating, you will get a whole host of issues. 

Thanks spritzer. I'm working on it now. Let me check.

[Pars]

For both Jose and penmarker: if you are only using an ss dynahi in SE mode, why even populate the - channels? All they are doing is burning power...
Even then, the - inputs would be tied to ground, as done in the original dynahi /dynalo.


Sent from my iPhone using Tapatalk

[penmarker]

33 minutes ago, Pars said:

For both Jose and penmarker: if you are only using an ss dynahi in SE mode, why even populate the - channels? All they are doing is burning power...
Even then, the - inputs would be tied to ground, as done in the original dynahi /dynalo.


Sent from my iPhone using Tapatalk

Having both SE and balanced input+output is good for versatility. A lot of headphones has non removable cables to get replacement XLR cables. For my case, my Matrix Mini-i DAC has balanced output but my turntable and CDP only has SE so I'll need that capability too.

For inputs I think it'll be okay since I'm planning to get balanced relay switches, but for the output should both LR ground and -ve be terminated to the headphone ground? Or how should it be? Only both channel ground without -ve connection?

[jose]

Thanks for the answer boys

Yes, I also think that having SE and balanced inputs brings a lot of versatility.

In my Stax amps I always bypass the cold channel (I use input selectors with relés) in SE inputs, but in the dynahi I hadn't through necessary... remember that I heard the crackling noise yesterday and I have this amp for a long time.

For the outputs, SE and balanced is also necessary by the different types of headphones that I have.

Yes, I understand that it's a waste of energy and that even results in the longevity of the components but we aren't talking about using a Grado. In my case use my Dynahi with high impedance and/or an disgusting sensitivity.

[Pars]

but for the output should both LR ground and -ve be terminated to the headphone ground?

Only the input - channel should be tied to ground, NEVER the output - channels. In effect you would be shorting the output devices to ground which probably wouldn't make them very happy:)

 

For SE output only the + signals and signal ground should be used.

 

That said, on my SS Dynalo Mk2, I have both RCAs and XLR inputs, with a toggle switch to tie the - inputs to ground for SE input via the RCAs. I don't hear a difference when using SE input and unshorting the - inputs via the switch. Maybe I need to listen more closely?

 

 

Sent from my iPhone using Tapatalk

[jose]

I did not remember that my input board bridges the Cold channel with ground when I select the a SE channel.

So that's not the problem.

 

[penmarker]

9 hours ago, jose said:

[snip]
I heard the crackling noise yesterday and I have this amp for a long time.
[snip]

I'm inexperienced with builds but can I just check with you how hot are the heatsinks? Are all of the output transistors fastened tightly against the heatsink? Do you have proper thermal paste/heatsinking compound?

6 hours ago, Pars said:

Only the input - channel should be tied to ground, NEVER the output - channels. In effect you would be shorting the output devices to ground which probably wouldn't make them very happy:)

For SE output only the + signals and signal ground should be used.

That said, on my SS Dynalo Mk2, I have both RCAs and XLR inputs, with a toggle switch to tie the - inputs to ground for SE input via the RCAs. I don't hear a difference when using SE input and unshorting the - inputs via the switch. Maybe I need to listen more closely?

Thank you very much this is very clear. 

I've had some ideas about the input/output wiring, so I needed confirmation. My personal take regarding inaudible changes is if you need to focus hard to detect changes then it probably wouldn't matter. But I don't know whether there is something happening electrically that could affect the unit in the long run, I doubt so.

 

Just last night some of my friends were having a group chat on our phones and they were discussing about resistor rolling. One guy showed tantalum resistors that cost around €5-7 a pop. He wants to upgrade his Cambridge Audio phono pre. I think that was a little mad, resistors don't change the sound.

do they?

But then again, I do see a lot of people using the same Dale resistors for the Dynahi.

[gepardcv]

2 hours ago, penmarker said:

But then again, I do see a lot of people using the same Dale resistors for the Dynahi.

That's because RN60D resistors have (1) fairly high-wattage ratings for their size, (2) high availability in many values, and (3) low cost.

[penmarker]

21 minutes ago, gepardcv said:

That's because RN60D resistors have (1) fairly high-wattage ratings for their size, (2) high availability in many values, and (3) low cost.

Ah ok that's enlightening. Here where I'm from, Multicomp resistors are cheaper compared to Dale at similar tolerance, value, and rating. But not much, maybe only 3/4 or 2/3 the price. We're talking cents here not dollars.

[rumina]

On 16.10.2017 at 12:32 PM, jose said:

Yesterday I was listening to my amp and I noticed a small sound coming from both channels, it's not constant and it's not perceived while listening music. It's a sound like "fried".

I was listening SE input and SE output and I don't bridge the cold input to ground.

I have been taking some measurements and BIAS is symmetric on both sides; 750/760 mVolt on R channel and 720/735 mVolt on L channel. On the contrary, offset drift a lot and I've never been able to keep it on 0 volt. It's always between -100~40 mVolt and 120~60 mVolt (without servo). With servo, the offset never exceed 20 mVolt.

It has always seemed to me that Mosfets run too hot for an on-board assembly (Mosfet reach 90°) although I know that the maximum temperature is 150° but even so, components and the PCB also get very hot.

I'm not sure where to start since I've changed the THAT and OPAs too and I still have the same noise.

Can Mosfet be damaged? Or can the problem be elsewhere?

Thanks

hi jose

do you have onboard heatsinks? i think the right bias for the onboard heatsink is around 0.375v but always only as much the amp get in a stable state (heatsink rise max 25 c over ambient or at cold places 30 c after 1 1/2 - 2 hours). if you only test the amp for 1/2 hour the amp can run away later and as the temps rise the bias get faster higher to a point the parts can fail/get damage. its a even bigger problem if the transistors are not well matched, the unbalance get fast out of a stable state, the fun and tricky part of a lot of nelson pass amp designs. 

i had some similar effects with a amp, some mosfet get to much current/heat but not failed completely, they made some strange chirr crackling noises (allways, sometime louder). was tricky because they measured fine whitout high current. what kind of noise do you hear? more static noise (as pink noise), hum or chirr/crackling noises? i think "fried" sounds like damaged but operational parts.

static noise is often because high gain and not so good matched parts. hum noise often fron grounding (wiring) problems (both channels same hum) or magnetic inducted noise from the transformator (the channel near the trafos hums louder then the other or only one channel has hum). 

hope helps a bit.

[jose]

Thanks Rumina.

I have reviewed my documentation but only find the reference at 0.75 Volt but  I'm regulating the Bias at 0.375 Volt and in a while I can tell something.

 

[jose]

Bias at 370/365mVolt. 

Offset is most stable,  arround -10~30 mVolt.

Now I only have noise on R channel. Noise is like a crackling and you can hear very low . With bal or SE I have the same kind of noise and with the same level. I'm sure it's not a ground noise.

 

[Pars]

Hmmm. I would probably start by swapping the input wiring (pot to boards). If the noise moves then it has something to do with that. Otherwise go over the board (components and location correct? Solder joints?)


Sent from my iPhone using Tapatalk

[jose]

I´m working on input wires and tomorrow I check both boards in search of some problem with solders. I wish it were just that!

The sound is so low that my oscilloscope can not detect it so I can´t show you any measure. 

However, at the moment I am also looking for a couple of boards (see 1.4 on board) in case I have to replace any.