A New Take on the Classic Pass Labs D1 with an ESS Dac

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi Guys,

I started a discussion with Acko, and it looks like everything should be a go! Once I get details on how the output section of his DAC is setup (layout, size, pinouts, etc...) then I can lay out the I/V stage and get boards made that will fit like a glove with the DAC.
excellent!!! now this is what i'm talkin' about!!

In the meantime I'll be tweaking the circuit a little more, and hopefully I can get some SE measurements this weekend.
i'm sure there will be some happy people right there.

qusp:

I ran the headphone option through the simulator, and it doesn't look like it's going to work. I'm not sure what the headphones need for voltage to get decent listening levels, but the output voltage has dropped quite a bit by the time you get a 300 ohm load across the output. You'd be down to under 2V p-p which may or may not be enough. I don't know how much voltage a pair of headphones would need.
well, I have a variety of headphones, actually my primary ones are only 28ohms, I use 680R in series to get just under 1vrms for them and that drives them to bleeding levels, the 300R headphones work fine without any resistors, I guess using the impedance of the headphones themselves to create the voltage needed. a simple unity gain diamond buffer is enough to drive most of my headphones

What I can do is add a connector at the output to allow a separate headphone amp to piggy-back on the outputs. If it were me, I'd probably build a scaled-down SOZ designed to drive headphones, but an op-amp could work well too. You could also use a multitude of other headphone amps designed specifically for the purpose and get very good performance. I often listen to music on my headphones at work, so I could be easily persuaded to do a headphone amp option.
oh yeah I know, there are plenty of other options for amping them, just the direct nature of driving direct from the DAC appeals, interested if you come up with something, but it would have to be balanced; all my headphones are recabled to balanced terminations. 2 of them have removeable cables that I could use an SE cale with, but I would prefer not to. I reckon we just see how it shapes up as is, but sure if you can think oif a good optional add-on, great. I already have a couple of separate headphone amps, but they are all single ended. i'm playing with some balanced designs of my own with the discrete buffers as output, so looks like i'll be sorted either way. there is a tube amp for them on the horizion too in the near future.

anyway mate, looking forward to this project even more now, I reckon we've got a serious contender here.

cheers

Jeremy
 
one thing; when I made my order from sumR for my iron, I totally didnt see this coming, so I only have 20va primary, with dual 15v secondaries. I might be able to eek that out to 18v DC in the regs, but it would be touch and go. is 15v DC enough to run the rails on the B1?? or is 18V the minimum?? obviously I could put the secondaries in series, but 12v is a lot to burn off in the regs, would need some serious sinks, could burn some off in a prereg I guess, but then I lose 2 secondaries. unfortunately its too late to change my order for the iron, they should be nearly finished by now; last update I got tyhey were up to enamelling.
 

opc

Member
Joined 2004
Paid Member
Hi qusp,

I have the exact same problem. The blue potted transformer I'm using now has 15V dual secondaries, and it's too close for comfort. I'm right on the edge of the regulator's dropout voltage. It does work though, but I wouldn't suggest it in the long term.

I ran a quick simulation with +/- 15V and it works just fine. You'll have less voltage swing capability, but I don't see why anyone would want 10VRMS out of a DAC anyway.

All you need to do is adjust the top resistors to get about 1/2 Vcc (7.5V) at the drain of each mosfet. In my sim, it was 180 ohms in place of the 200.

The adjustable gain feature is pretty slick, since all you need to do is increase the 200 and 400 ohm resistors. I personally don't like a lot of gain, since most systems have far too much as it is, and it results in more noise and distortion for no reason. With this circuit, you get to balance gain with the amount of voltage swing you're willing to accept at the output of the DAC. Less gain = less voltage swing at the DAC. I think a nice 2VRMS at FS is a good way to go, but if someone wants more, it's easily done.

One of the things I want to try is dropping the rails down as low as possible, and seeing at what point it negatively affects the sound and the performance. If you only want 2VRMS output, then the rails should be able to go much much lower which would result in less dissipation on the parts, and more flexibility in the power supply. I kept them high in order to allow people to get a lot of gain, but maybe I'll make it an option.

Anyone out there with an opinion on gain? Anyone who wants more than 2VRMS at 0dBFS? If so, can you tell me why, and how much you would want?

Cheers,
Owen
 
Hi Owen

I have the exact same problem. The blue potted transformer I'm using now has 15V dual secondaries, and it's too close for comfort. I'm right on the edge of the regulator's dropout voltage. It does work though, but I wouldn't suggest it in the long term.

ahh I see, well at least it will work initially. so are you saying you are driving the D1B1 with the one potted trannie with 2 x 15v secondaries?? or just mentioning the spec as I was. I have 2 x fully encapsulated cans with a single 240v/20va primary and 2 x 15v secondaries for the VA section and 1 x 20va/2 x 10v for digital does this improve matters?? likely not I guess, still the same output, but not as strained as if it were just the one.

I ran a quick simulation with +/- 15V and it works just fine. You'll have less voltage swing capability, but I don't see why anyone would want 10VRMS out of a DAC anyway.
nah I dont need more than 2vrms for any reason I can think of. as I said the only things i'm wanting to drive are headphones and biamped powered monitors. only thing I can see in that future is clipping.

All you need to do is adjust the top resistors to get about 1/2 Vcc (7.5V) at the drain of each mosfet. In my sim, it was 180 ohms in place of the 200.
ok, nice one, looks like I have another reason to make another order with texas, perhaps a GB would be in order for TX2575 for the gain setting resistors?? but thats something to talk about later. I cant fill the B1 with them $$$$, but if you can mention any other specific positions that you have experienced particular benefit from really tight tolerances and matching, i'll grab some for those slots in 0.005% ;) love the texas components R's

The adjustable gain feature is pretty slick, since all you need to do is increase the 200 and 400 ohm resistors. I personally don't like a lot of gain, since most systems have far too much as it is, and it results in more noise and distortion for no reason. With this circuit, you get to balance gain with the amount of voltage swing you're willing to accept at the output of the DAC. Less gain = less voltage swing at the DAC. I think a nice 2VRMS at FS is a good way to go, but if someone wants more, it's easily done.

that is pretty sweet and at any time to be able to adjust it with a coupe of parts is great. but yeah the standard line level 2vrms is good by me, leaves it flexible what gear it can be used with without a pre/. dont hold me to that just at the minute, but at this stage I cant see any reason for me going higher. i'll have a think on it. plus if others pipe up with reasons why, i'm all ears

One of the things I want to try is dropping the rails down as low as possible, and seeing at what point it negatively affects the sound and the performance. If you only want 2VRMS output, then the rails should be able to go much much lower which would result in less dissipation on the parts, and more flexibility in the power supply. I kept them high in order to allow people to get a lot of gain, but maybe I'll make it an option.

Anyone out there with an opinion on gain? Anyone who wants more than 2VRMS at 0dBFS? If so, can you tell me why, and how much you would want?

Cheers,
Owen

yeah I mean we arent talking DNR here, if so then I would be all for it, but more gain in this circuit means you pay somewhere. I would definitely like it as an option, in the meantime I will keep researchin a partner for this dac in the next stage. speakers other than my monitors are out of the frame, so it would only be if I were to get better performance from my amp with more gain. but so far it seems llike it will suit my needs pretty well without one for the moment.
thanks for the numbers re the gain setting resistors. great dialogue so far. keen as mustard

cheers

Jeremy
 
Last edited:
I thought the DAC chip like all S-Deltas had 1/2Vcc DC offset and the D1 stage was designed for R2R DAC with 0 offset, I don't see this being dealt with in the circuit, something to check.

Also we really do need to have a nice discrete balanced to unbalanced mixer to reduce common mode noise and let the DAC chip show its full potential.

Very nice project btw.
 

opc

Member
Joined 2004
Paid Member
Slow over the Christmas Break

Hi Regal,

I've addressed this a few times, and I think I mentioned it in the first post. The normal D1 uses trimpots to set the gate voltage on the mosfets. This in turn is used to set the voltage at the source to 0. In my circuit, the same thing is done, except the voltage at the source of the mosfets is set to 1.65VDC which is 1/2 VCC, exactly what the DAC wants to see there.

The top two resistors feeding the drains of the fets are then adjusted to get about 9V at the drain of the mosfets, which is what the B1 wants to see at it's input. The final current through the mosfets is about 50mA, which is much much higher than the D1.

I have the exact values for all the resistors in the working circuit, but I'm out of town for the holidays and I won't get a chance to work on this again until I get back on the 29th. From what I can remember, you need 400R on the sources, 180R on the drains and adjust for 1.65V at the source.

You are correct about the balanced to single ended converter. This circuit works with a single ended output, but you lose all the advantages of the balanced DAC and the balanced nature of the D1B1. Performance definitely suffers for it.

I personally don't care to have SE outputs since the entire rest of my system is balanced, but I can understand the need for a good SE output.

Anyone with any suggestions? This might be a good spot for a single UG op-amp with good CMRR. A transformer would also work very well here.

Happy Holidays!

Owen
 
Hi Regal,

I've addressed this a few times, and I think I mentioned it in the first post. The normal D1 uses trimpots to set the gate voltage on the mosfets. This in turn is used to set the voltage at the source to 0. In my circuit, the same thing is done, except the voltage at the source of the mosfets is set to 1.65VDC which is 1/2 VCC, exactly what the DAC wants to see there.

The top two resistors feeding the drains of the fets are then adjusted to get about 9V at the drain of the mosfets, which is what the B1 wants to see at it's input. The final current through the mosfets is about 50mA, which is much much higher than the D1.

I


Thanks, I just don't see how you can come up with the 1.65V at the Mofset Source from the datasheet, running 50 mA and all the charts are in amps. I guess you need Spice or something.


It would be very useful for a lot of us with 5V DAC's to know how to change R1&R3 to get 2.5V at the source.
 

opc

Member
Joined 2004
Paid Member
Hi Regal,

Sorry for the late response. I've attached a schematic that will work well with DAC's requiring 2.5V at the output.

Ignore the 33k3 value for R11 and R12, what you need there is a 50k pot. Start about midway (25K) and adjust to obtain 2.5VDC at the source of the fet. When that is set, you should see very close to 9VDC at the drain.

I'm not sure how much gain you need, but if you tell me what DAC and what configuration you plan to run, then I can dial in R1, R2, R3 and R4 to get you whatever you want voltage-wise at the output.

This project has been put aside for the past few months, but I should be getting back to it shortly.

Cheers,
Owen
 

Attachments

  • DAC.JPG
    DAC.JPG
    86.1 KB · Views: 2,089

opc

Member
Joined 2004
Paid Member
Hi Guys,

I'm about to take a second run at the layout, and I have a general question about what people prefer as far as SMD and PTH components go.

I've been building with SMD parts for a while now, and I definitely prefer them, but I have noticed that people seem to prefer building with though-hole parts.

What is the general consensus on parts for a project like this? I've got some preferences of my own, and maybe people would like to comment:

Refer to the schematic attached above for designators.

R1, R2, R3, R4 - Panasonic 3W metal oxide or Caddock MP930 or 4x 1W thick film SMD (2512)

R5, R6, R7, R8, R9, R10, R13, R14 - Standard 1/4W PTH (300 spacing) or 0805 SMD

C3, C4 - 5mm spacing 10mm diameter solid polymer Nichicon - PTH

C1, C2, C7, C8 - 0805 SMD C0G/NP0 5% or PTH C0G/NP0 5% or PTH film

R11, R12 - Pots - standard 3-pin through-hole?

C5, C6 - Space for large film caps

J1, J2, J3, J4 - dual LSK170 SMD SOT123 or single LSK389 SMD SOIC

I could probably provide the boards with SMD parts pre-soldered for those who are afraid, but overall, it's easier than PTH to work with.

Any comments would be great!

Cheers,
Owen
 
Hi Owen,

I have search the forum and it seems you are the only guy who actually use a different mosfet with the original (almost) D1 circuit.
I order a few different mosfet to try as well (I can't find the RJK0305.
Is there any concern with the high input/output capacitance of this fet.
Also, I have notice that the transconductance drops lile a rock with low current for a lot of high transconductance device, how did u pick this one out from the bunch? Did u compare the sound between this and the original irf610? Thanks.

...chuck


Hi Guys,

..

I've combined the excellent transconductance of a power mosfet, with the low distortion of a JFET buffer to get to the circuit below. It's basically a low impedance D1, with a different mosfet, and a B1 buffer in place of the old mosfet based buffer. You could probably call it a D1B1.

..
Cheers,
Owen
 

opc

Member
Joined 2004
Paid Member
Hi Pidesd,

I suppose both congrats and condolences are in order. Could you draw up exactly the circuit you were using, and what the voltages were at each of the fet nodes with respect to ground?

I've only tried the one mosfet so far, but I took the time to simulate several of them, and there wasn't much difference between them other than the voltage required at the gate to adjust the circuit.

I should be clear about power dissipation and requirements:

- R3 and R4 dissipate just under 1/2 watt each. Use at least 1W resistors.
- R1 and R2 dissipate just under 1 watt each, use at least 2W but 3W is better.
- Each mosfet dissipates a little over 1/3 of a watt, so it will be toasty.
- All other parts can be anything you want.

If you're uncomfortable running the fet that hot, then either add a little heatsink (the small tab mount TO-220 should do the trick) or use SMD parts and add some copper area to the PCB.

I'm glad you took the time to build it, and if you post your schematic, I'm sure we can figure it out!


Chuck,

You are absolutely correct about the transconductance of the fet dropping drastically at low current, and after quite a few simulations, it looks like the fet plays a smaller role than I initially assumed. I need to verify this with real tests, so I'll be doing that over next week. If anyone has any favorites, let me know and I'll include them in the test. Here's the short list:

IRF510
IRF610
IRF640
FQA19N20C
FDP030N06

As for input capacitance, I don't think it will have much of an effect, since the gate doesn't swing in this application. High Coss and Crss might have a negative impact, but I doubt it. I'll have to check. Keep in mind the conditions for all these specs in the datasheet, and that they are quite a bit different from what we're doing with it. We're running it at 50mA with only 7.3Vds. The swing at the source is less than a few mV and at the drain it's only a few volts.

Also, remember that the distortion and other performance specs on page 1 were obtained with the little Renesas fet, which is the only fet I've tested. I'll test the others mentioned above and post a comparison.

Cheers,
Owen
 
Last edited:
Hi Regal,

Sorry for the late response. I've attached a schematic that will work well with DAC's requiring 2.5V at the output.

Ignore the 33k3 value for R11 and R12, what you need there is a 50k pot. Start about midway (25K) and adjust to obtain 2.5VDC at the source of the fet. When that is set, you should see very close to 9VDC at the drain.

I'm not sure how much gain you need, but if you tell me what DAC and what configuration you plan to run, then I can dial in R1, R2, R3 and R4 to get you whatever you want voltage-wise at the output.

This project has been put aside for the past few months, but I should be getting back to it shortly.

Cheers,
Owen

Thankyou, probably would be the TI PCM 1794, 7.8 ma p-p
 

opc

Member
Joined 2004
Paid Member
I forgot to give the update on the layout... everything is pretty much done, and I'm about to send it off to Acko to check against his PCB. There are some differences, but I'm going to see where he can give a little, and where I can.

As you can see, it's not a lot of space (100mm x 72mm), but everything seems to fit. I've crammed the +/- 18V regulators on the board, but it's a very tight fit, and I'm not sure if people would rather have the regs off-board. Any opinions are welcome at this point. I personally would rather have them on the board, and I can provide tie-points for bypassing them if you want to use your own off-board regs.

I still need to talk to Acko about running the board wider than his DAC, which I don't think will be a problem. If that's a go, then I can layout the regs in a more optimal manner.

The board is two layer, ground plane on the bottom, signal and power on the top. I'll be splitting the ground plane in the middle of the board, so each channel has it's own plane. I'll provide multiple tie points for grounds, giving the user the option of tying the grounds at the DAC, at the PSU, or on this board itself.

The only other issue is output coupling caps. I've left space for 27.5mm spaced radial caps, that are 18mm wide and 31mm long. I'll be using 8.2uF Epcos polypropylenes (B32674D3825K) but others may want to use their favorite caps. As you can see, I've already given up over half the board area to caps, and I can't afford any more. I can provide pads for smaller caps, as well as electrolytics if people would like, but you're not going to be able to put monster sized films in those spots unless you mount them off-board and wire them over with flying leads.

Check it out and let me know if there are any comments!

Cheers,
Owen
 

Attachments

  • D1B1.pdf
    15 KB · Views: 414
  • D1B1.jpg
    D1B1.jpg
    132.9 KB · Views: 1,892
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.