KG Balanced Dynahi build discussion thread

[penmarker]

The heatsinks are pretty hot with 750mV, plus I'm using a Sigma22 for both boards and their regs are heatsinked to the amp floor panel. The PSU floor is hot too with one board at 750mV so I've lowered it to only 450mV for now.

I haven't tried it out yet as my balanced 10k pot it still in the mail. I was wondering will there be any adverse effects if I use my balanced DAC (Matrix Mini-i) as pre out to the dynahi without the 10k pot? Can't really find the output impedance for the Mini-i XLR at the back, but the headphone out is 12 ohms.

[penmarker]

On 11/5/2016 at 8:52 AM, Pars said:

This is a balanced Dynahi? That is one measurement (offset of O+ to O-). First, you should set DC offset for O+ to gnd and O- to gnd. Get these both as close as possible; should be able to get each to within 5mV without the servo. If you can't, get them as close as possible. Then check O+ to O-. Not sure what these boards have for pots on them. By going back and forth, you should be able to get them all pretty close.

EDIT: and have the inputs either grounded or turn the volume pot all the way down.

So I am measuring the offset for my balanced build and the O+/O- has near 0mV offset, but the O+/gnd and O-/gnd has 3.3V offset. Is this correct? Won't this destroy single ended headphones since their ground is tied?

[jose]

If you have 3.3volt, something is wrong.

[Pars]

They should all be 0V, or very close. Ideally ~1mV or so without the servo. They will drift, so more realistically between +/-2mV or so, O+ and O- to Gnd. Inputs grounded or pot at 0.

 

EDIT: looking at your picture above, what are you using to isolate the outputs from the heatsink bracket? Thermal TO220 pads are best for this; I would not depend on thermal compound for this.

 

Also, it does not look like you have insulators under the screws for each device? The tabs on these are live (collector of each device) and MUST be insulated; whether you use nylon or PEEK screws, or steel screws like you are, but with the nylon step insulating washers.

Sent from my iPhone using Tapatalk

 

 

[penmarker]

Jeeybus, I'm fucked. I can't lower the O+/gnd O-/gnd 3.3V offset at all.

[Pars]

What does V+ and V- look like? Also, with the amp off, set your meter to ohms and check continuity from each output device tab (tab, not the screw) to the heatsink. Also, check from a tab on one of the first 4 devices on the left to the tab of one of the 4 devices on the right side.

 

 

Sent from my iPhone using Tapatalk

 

 

[jose]

Have you tried to download the BIAS? 

[penmarker]

11 hours ago, Pars said:

What does V+ and V- look like? Also, with the amp off, set your meter to ohms and check continuity from each output device tab (tab, not the screw) to the heatsink. Also, check from a tab on one of the first 4 devices on the left to the tab of one of the 4 devices on the right side.

V+ and V- look okay, 29.5V and -29.87V.

I've checked continuity when I mount the devices to the L bracket heatsinks. Yes there's mica insulators with plastic washers but I cut down the mica heatsinks to size because they're too wide or else they won't stay flat and stack up. That could be the point of failure there.

I check continuity on the tab after every 4 devices mounted. I didn't check cross continuity between the devices on the right and left however, I checked continuity on random output device against the heatsink because I assumed if one has continuity then all of them will have continuity.

The first 4 devices on the left to the tab of one of the 4 devices on the right side should NOT have any continuity correct?

6 hours ago, jose said:

Have you tried to download the BIAS? 

I don't understand you.

[Pars]

The PNP and NPN should not show low resistance (i.e., short) to each other. They will probably show some resistance, but I would assume it would be in the K ohm or higher range. None of the tabs should show a short to the L bracket. I couldn't tell in your pic what was under there or whether you had the shoulder insulating washers (such as Mouser 532-7721-7PPS). If the N and P were shorted together, you probably wouldn't be seeing approximately correct V+/V- (assuming the PSU was connected to the amp boards when you took your measurements), as they would be connected to one another via the shorted collectors.

Just a shot from looking at the pics.

EDIT: and I think Jose was asking if you had decreased the bias. Personally, IMHO, this amp is designed to run at 750mV/75mA bias, so that is what you should be running, and it shouldn't effect offset (i.e., you should be able to get the offset down into the sub 5mV range at that bias).

Edited March 27, 2018 by Pars

[jose]

Yes, only like a test, decreased Bias for example until 600mVolt and check the offset.

[jose]

Have you tried to change the THAT340 for other?

[penmarker]

5 hours ago, Pars said:

The PNP and NPN should not show low resistance (i.e., short) to each other. They will probably show some resistance, but I would assume it would be in the K ohm or higher range. None of the tabs should show a short to the L bracket. I couldn't tell in your pic what was under there or whether you had the shoulder insulating washers (such as Mouser 532-7721-7PPS). If the N and P were shorted together, you probably wouldn't be seeing approximately correct V+/V- (assuming the PSU was connected to the amp boards when you took your measurements), as they would be connected to one another via the shorted collectors.

Just a shot from looking at the pics.

EDIT: and I think Jose was asking if you had decreased the bias. Personally, IMHO, this amp is designed to run at 750mV/75mA bias, so that is what you should be running, and it shouldn't effect offset (i.e., you should be able to get the offset down into the sub 5mV range at that bias).

I'll test that when I have the time. Here's a closeup pic. You can see (just barely) there's an insulator and a washer to insulate MJE devices from the L bracket.

For sanity check I've double/triple checked pin orientation with respect to the silkscreen and all components correspond to the silkscreen orientation (had this problem with a NAP200 build where one of the transistors are backwards).

32 minutes ago, jose said:

Have you tried to change the THAT340 for other?

I only have 2 units and both boards are exhibiting same behaviour. I tested one of the THAT340 prior to installation, it measures fine and matched within respective P-N channel.

 

So the MJE are rated up to 150V while the MPSW are rated up to 80V. How likely have I fried the devices on my board? Or is it current that fries them and not voltage?
There's no smoke, no smell.

[Pars]

Doubtful that you fried any devices unless you connected the power backwards.

1. When you say you replaced resistors with 340R, these were the 500R resistors?
2. And those clear LEDs, have you verified their forward voltage as 1.7V with a meter?
3. No PNPs where an NPN should be and vice versa?
4. All resistors ohm out as their marked value?

Edited March 27, 2018 by Pars

[penmarker]

18 minutes ago, Pars said:

Doubtful that you fired any devices unless you connected the power backwards.

1. When you say you replaced resistors with 340R, these were the 500R resistors?
2. And those clear LEDs, have you verified their forward voltage as 1.7V with a meter?
3. No PNPs where an NPN should be and vice versa?
4. All resistors ohm out as their marked value?

Thanks a lot for the checklist pars.

1. Yes

2. Didn't verify by meter but their spec is 1.7V. http://my.element14.com/webapp/wcs/stores/servlet/ProductDisplay?catalogId=15001&langId=83&urlRequestType=Base&partNumber=2062468&storeId=10188

3. Yes.

4. I didn't verify by meter one by one but their packaging and color bands check out. 1% so won't stray too far.

[Pars]

1. OK
2. I would measure them. Typ is 1.7V; max is 2.2V. We used to match LEDs for these, more important back when there weren't pots for offset.
3. OK
4. I've seen or heard of resistors being mismarked. Since you have the same problem on both boards, unlikely however (unless it is the whole batch of one value).

[mwl168]

No sure this applies to this situation, but are the inputs shorted to ground when you ran your tests and adjustments?

[jose]

1 hour ago, mwl168 said:

No sure this applies to this situation, but are the inputs shorted to ground when you ran your tests and adjustments?

Good observation

[penmarker]

Nope nope guys shit. Sorry I feel so dumb. My biasing technique was wrong. The offset is fine. It goes below 40mV easily and the servo lowers it down to 0mV and maybe fluctuates a bit between 4 to -4mV

[Pars]

What was wrong in your biasing technique? Is this with the replaced resistors? What is bias set at now?


Sent from my iPhone using Tapatalk

[penmarker]

What I did was I turned up the two pots beside the 500R together without looking at the offset properly. So it just kept going up while I was trying to focus counting to match both pots' rotation. To correct the offset I had to turn the other one less or more and that fixed the offset right up.

Resistors are different now; 50R are now 100R and 500R now are 680R. 

[penmarker]

Dude.... this sounds amazing. Miles away from the Lehmann headphone amp.

Should I worry about the initial startup offset? Its peaks to about 400mV maybe, but into my 150 ohm headphones that's like 0.0027 watts.

 

**edit: Goddamn can't listen to binaural tracks now, its 3.27am and I'm alone in the house and its just to creepy to hear all these sounds around me with nobody in sight.

Edited March 28, 2018 by penmarker

[jose]

It's good to know that it sounds. 

As for the peaks, I don´t like to turn on the amplifier with the headphones connected to it. I prefer turn on and wait a few seconds... but I also know people who don´t like to connect the headphones with the amplifier turned on by the short that the connector produces 

[Pars]

It’s always a good idea with these to start with the pots centered and incrementally adjust the offset out, then set the bias (IIRC correctly the adjustment procedure). You can get the pots approximately centered by using an ohmmeter and getting both sides to read approximately the same wiper to end terminal.


Sent from my iPhone using Tapatalk

[kevin gilmore]

you can always add the protection circuit board.

 

[penmarker]

There's no turn on sound but turning it off does have a soft thud. Don't think I got space in the chassis for a protection board. If I'm to revisit this I'll make a dual chassis build. Or probably get a large one for speaker amps.

Unfortunately I got the pass clone chassis with an 8mm thick front panel. I'll figure something out for the volume pot and channel selector.

The whole thing is hot but I'm sure it's fine. My thermometer is in the mail, so I'm gonna take note of the temps later.