A version of an O2 Desktop Amp (ODA)

Everyone who expressed interest previously in these extra boards has contacted me now, so all the boards are spoken for at this point. I may do another run at some time down the road.

However by using that 4 layer gerber file I posted a few posts back in this thread, anyone can do their own run at any time! 5 4-layer boards at Seeed Studio (where I in China had it fabbed) is about $100 as I recall plus shipping. Slow boat shipping (probably 1 - 2 months with no tracking) is free, DHL (3-4 days with tracking) for 5 was around $20. 10 boards was $120 plus $30 DHL. I'm still amazed at those low prices for 4-layer boards, along with the good quality of what came back.
 
Last edited:
Thank you very much for your detailed response! And I apologize, it seems I haven't read the thread carefully enough (or clicked an old schematic) because I completely missed that you swapped the 627s for the 49990s! :)

Hey I'm glad you asked the questions! I'll bet several other folks stopped reading the thread after those initial OPA627 experiments. I also came to learn that the OPA627 has been around a long time and a great many other OPA627-based headamps are out there. The main point in using the OPA627 in the initial design was to put the pot in front, due to the low input bias current of the FET amp. But with the pot back in the middle it doesn't matter anymore. Might as well use a lower-noise, lower distortion, and cheaper bipolar input chip like the LME49990. I putzed around with using the LME49720 but in the end the LME49990 has better AC specs and 2x is the same cost as the one dual LME49720.

One interesting thing, I eventually found that going to the 0.5R balancing resistors on the parallel output stage (vs. the 1R resistors in the O2) completely eliminated the 80kHz oscillation. This is further down the thread when I was using the OPA627 wrapped around the output buffers as a DC servo. I ran into that same 80kHz oscillation and the 0.5R's solved it. I believe the issue is shown in the upper left figure of page 11 on this Linear Tech data sheet:

http://cds.linear.com/docs/en/datasheet/1010fe.pdf

They have the feedback loop taken off before the balancing resistor on just one of the parallel output current buffers. The other parallel buffers are then just slaves in a "best effort" arrangement. RocketScientist / NwAvGuy warned against taking feedback off after the balancing resistors, noting that would essentially add current feedback to the voltage feedback from the IxR drop (in the 1R reseistors) caused by the output current. That is one reason he said that he chose a feedback-less design for the O2 output stage.

It seems that effect was causing the 80kHz oscillation when using the 1R balancing resistors. Dropping them to 0.5R became "close enough" to no resistor so solve the problem. It would be extremely interesting to try 0.5R balancing resistors in the O2 amplifier instead of the 1R's and run RocketScientist's suite of dScope measurements again (massless are you out there? :)) to see if they are better, worse, or the same.

I did eventually find a source of 0.5R 1/8W metal film through-hole resistors for this project here on eBay from Hong Kong. I can't get a link to work for some reason, but search eBay for "0.5 ohm 1/8W". $3.99 for 100 including shipping. Neither Mouser nor Digikey has 0.5R in 1/8W. In the BOM I've had to specify 1/4W 0.5R that has to be mounted on end. These 1/8 watters fit right in and would save some money on the build.

On the resistors, I re-read what I wrote above and see several typos. Argh. :) For anyone else reading this with the same questions, it should read that thin film is lower noise than thick film with the SMD stuff. I used thin film for the signal path parts (usually around $0.80 each) but went back and substituted thick film (usually around $0.10) for parts in the relay circuit and clipping circuit to cut some costs.
 
Last edited:
In the figure you mentioned, the output are not in parallel.
Each "slave" is driving it's own, separate output.

You are right! Thanks for the heads up. I went fishing on Google for the diagram I had in mind and thought I had found it. Whoops. :) It would be that same diagram, but with those two output balancing resistors connected together to form one channel.
 
Here are the latest build instructions for this project. Please let me know about any typos or anything that isn't clear and/or needs more explanation. :)

agdr - I'm planning to make a small run of PCB's using your gerber files for personal use. Looks like a cool project to build. Just curious if you plan on making any more revisions to the v1.4 gerbers?

Thanks for your contributions.
 
2000impreza - Hmmm... if you are doing another run there may be a couple of minor cosmetic things that may be worth doing.

While I was doing the build I was thinking it would be nice to add "NEG" lettering on the negative size of the large electrolytics. The Eagle library has the standard "+" sign on the + lead, but the caps always have the negative side marked. I've never seen anyone do that with a PC board but it seems like a good idea that would be helpful during the build.

The board has vertical diodes in a few places and I have lettering that says "band up" or "band down". It might be nice to make the font on that one size larger too. A few more minor things like that which would be "safe" to change without doing another test board since they don't alter the functionality in any way.

So if you would be willing to wait I could probably have those done and posted by the first of next week.

One hiccup to be aware of on the build. You have to have a back panel with this project since the voltage regulators heat sink to it. I'm still in the process of sending out the panels and getting them back to verify so that could take a few more weeks, espcially if I mess something up in the CAD file and have to send them out a second time. I'll be posting measurements to hand drill it next week, so you can always drill the panels yourself that come with the B4-080 box. I hand drilled mine. I'm increasing the accuracy of the measurements though and should have those by next week to post. The back panel centers with the RCA jack mounting screw on the left and the power switch hole on the right. Then the voltage regulators mount on with mica washers and are soldered down.

Also note that the PC board goes into the second slot up from the bottom, not the bottom slot. That is because the B4-080 case has a panel screw and extrusion at the bottom of the case right in the center. The extrusion sticks up a bit. Using the second slot gives lots of clearance between the bottom of the board and the bottom of the case. All of the panel measurements I'm posting and the professional panel files for Proto Panel and Front Panel Express have the board in the second slot.
 
Last edited:
Panel hole dimensions

Here is the set of panel hole dimensions I just sent out to Proto Panel for the front and back panel. I won't know if there are any mistakes until I get the panels back so it is probably best to wait, if possible, but if someone is set up with a CNC system and some stock and wants to give it a try, let me know the results! :)

These panels have the lettering left out to keep the costs down. Once these measurements are verified I'll post the CAD files for both panels that their "Cetina" software uses. When placing your panel order it is easy enough to fire up the software and just add any size, font, etc. of lettering wanted. Lettering raises the price and I know from past projects is pretty user-specific. Everyone has their own favorites on fonts and what the labelling is.

I'll also post a set of Front Panel Express CAD files once these measurements are verified.
 

Attachments

  • ODA panel CAD dimensions 3.pdf
    64.3 KB · Views: 107
Last edited:
Gain switch out of stock, use 3 position

I've discovered that Mouser is now out of stock of the 4-position 2-pole rotary gain switches for this project, Mouser #105-SR10010F-24NS. I've just had a chat with them and Mouser is being good enough to order in another 1000 which should arrive sometime in September.

In the meantime the 3 position version of the same switch can be used, Mouser #105-SR10010F-23NS with 489 in stock. Fits in the same PC holes just fine, you just get one less gain switch position.
 
V1.6 4-layer Gerber files, layout, circuit

Here is one more update with a bunch of minor cosmetic updates I noticed during the build. This is still 4-layer, with the DRC and CAM set up for Seeed Studio's 4 layer fab service.

2000impreza - I had forgotten that V1.4 was the last 2 layer set of Gerber files. The 4-layer that was fabbed was V1.5 and this 4-layer is V1.6. Given how (relatively) cheap these 4-layer boards are to fab these days, I would definitely recommend going with the 4-layer version/files. I've updated the wording in post #1 of this thread to point to these latest V1.6 files.

Some of the changes:

  • "NEG" lettering added to the negative lead of the large electrolytics, along with the standard "+" on the positive lead.
  • Increased the font one size for the letting on the vertical diodes that says band down or band up, just to make it easier to read.
  • Decreased the hole size one notch on the DC power jack.
  • Increased the hole size one notch on the 3.5mm jacks. No problem here on the existing boards, but during the build I accidentily slopped some solder in one of the holes and had to suck it out with the solder sucker (solder wick works too) to get the jack in.
  • The silk screen outline for the 4.7uF coupling caps was square from the Eagle library while the part is 8.4 x 7.2. On the existing boards I left appropriate space between the caps to compensate. I've gone back and added the exact shape to the library.
  • The two SMD 0.001uF cap pads under the 4.7uFs are removed. These are not used in the build and are left over from my past DC servo attempts. I hadn't noticed they were still there in V1.5. Just leave those pads unused with the existing boards.
  • Added a SMD pad for a 10 ohm 1206 SMD R74 in series with R31 in the power supply section. Now this is getting really perfectionist, :p The rail runs at 7.020Vdc in the existing builds and that 10 ohm resistor moves it to 7.000Vdc. I had space so why not. You can just put a 0 ohm jumper in there or short it with a piece of wire and solder if you don't want to mess with it.
  • Added a 1206 SMD resistor pad in series with the output on each channel of the pre-amp chip, going to the RCA pre-amp outs. For very long RCA cables some chips may be happiest with 20R or so of series resistance to decouple cable capacitance. This can easily be added with the existing boards by cutting the trace between JP15 and J5 with an xacto knife and soldering in a through-hole resistor between the pins. If not using an output resistor just short those across with a 0R 1206 jumper (in the BOM) or a wire.
  • Moved R27 in the power section up slightly from the board edge. If someone left an extremely large solder ball there it could possibly have touched the back panel. No problem on the existing boards - if you use too much solder on that end of R27, just use some solder wick to suck a bit of it off.
  • Moved C37 and C41 to better take advantage of free space on the board.
 

Attachments

  • 80mm x 160mm 4 layer 1.6mm green HASL 100pct etest.zip
    206.9 KB · Views: 81
  • desktop headphone amp circuit V1.6.jpg
    desktop headphone amp circuit V1.6.jpg
    542.5 KB · Views: 664
  • desktop headphone amp layout V1.6.png
    desktop headphone amp layout V1.6.png
    738.6 KB · Views: 617
  • desktop headphone amp layout V1.6.pdf
    800.7 KB · Views: 106
  • desktop headphone amp circuit V1.6.pdf
    945.7 KB · Views: 182
Thanks for the update. I just sent off the updated 4 layer gerbers to seeed studio to run a batch of 10. The price difference between 5 and 10 was small enough I just went with 10.

For the 4 position rotary switch, ALPS SRBM140800 looks like it should work. Will take a closer look at the datasheet later. Mouser is out of stock but it is available from online components.

SRBM140800 Alps Electric | Rotary Switches SRBM140800 | Onlinecomponents.com
 
Last edited:
2000impreza - good find on that Alps switch! I had completely missed that one. From the data sheet it does look like an exact copy - or vise versa - of the Alpha switch. Mouser has 61 in stock of the version with the T18 15mm shaft (688-SRBM140700), which would match up with the 15mm shaft on the Bourns pot.

That Alps switch would actually solve a couple of problems. Right now the 20mm shaft on the Alpha switch sticks out 5mm past the shaft of the Bourns pot and may need to be cut, depending upon the knob being used. With a 15mm shaft it could use that same skinny knob from Mouser that I have in the photos on the pot. And related to that, the Alpha switch has an oddball D-shaft size. I couldn't find any of the pre-molded Mouser knobs that would fit (shaft slightly too big in diameter) and went to a knob with a setscrew. That Alps 140700 has the same T18 shaft as the Bourns pot that would fit the knob. I've attached the list of Mouser part number for that knob below in the various color combinations.

The 10 board option from Seeed is definitely the better cost-per-board. With DHL shipping included I think it works out to something like $24 a board for 5 boards vs. $15 a board for 10.
 

Attachments

  • knob.PNG
    knob.PNG
    81.1 KB · Views: 510
Last edited:
Thanks for the update. I just sent off the updated 4 layer gerbers to seeed studio to run a batch of 10. The price difference between 5 and 10 was small enough I just went with 10.

For the 4 position rotary switch, ALPS SRBM140800 looks like it should work. Will take a closer look at the datasheet later. Mouser is out of stock but it is available from online components.

SRBM140800 Alps Electric | Rotary Switches SRBM140800 | Onlinecomponents.com

2000impreza - are you going to use all your PCB'c, or would you consider selling one to me in Sweden? :)
I would love to compare this amp with my PPA.

I will probably order most of the the components from Mouser, and will try to find something to replace the DP4T ALPS switch with - it will probably be too expensive to order just one component from OnlineComponents.

---
Sent a long time ago from a Galaxy [Tab] far far away using Tapatalk HD
 
2000impreza - are you going to use all your PCB'c, or would you consider selling one to me in Sweden? :)
I would love to compare this amp with my PPA.

I will probably order most of the the components from Mouser, and will try to find something to replace the DP4T ALPS switch with - it will probably be too expensive to order just one component from OnlineComponents.

---
Sent a long time ago from a Galaxy [Tab] far far away using Tapatalk HD

Heffa - I definitely do not need all the pcb's. Realistically I will probably only use 2 or maybe 3 to experiment with. I have no problems with sending you a PCB. Before that happens I want make sure there are no obvious quality problems with the seeed boards and that I am not breaking any of the forum rules. Let me get back to you after I receive the batch of boards and get a chance to look them over.

Regarding the gain switch, agdr identified an alternative part (688-SRBM140700) that can be used. This part is currently in stock at Mouser.
 
Heffa - I definitely do not need all the pcb's. Realistically I will probably only use 2 or maybe 3 to experiment with. I have no problems with sending you a PCB. Before that happens I want make sure there are no obvious quality problems with the seeed boards and that I am not breaking any of the forum rules. Let me get back to you after I receive the batch of boards and get a chance to look them over.

Regarding the gain switch, agdr identified an alternative part (688-SRBM140700) that can be used. This part is currently in stock at Mouser.

That sounds great - I'm in no hurry, won't probably have time to build anything for some weeks/months anyway :)

---
Sent a long time ago from a Galaxy [Tab] far far away using Tapatalk HD
 
Panels are back

Well I have to say that I'm impressed with Proto Panel! The panels for this project are back already. I emailed the order just last Wednesday, so that is just 4 business days to get them in my hands. Wow. Disclaimer - I'm not affiliated with them in any way, just a happy customer at this point.

They honored the 40% off special they have running through the end of the month. For the panels with no lettering it worked out to be around $8.60 a panel with the discount, plus $6 shipping. Can't beat that. For panels with lettering it looks like it will be around $12 a panel with the discount.

The first photo below is the received panels. Packed in bubble wrap and also shrink-wrapped. The next is the back panel installed, and the third is the front panel.

The two CAD renderings are the panels with lettering. I'm going to send those in next to see how lettering works out. Proto Panel says they can only do engraved lettering for the small batches. For silk screened they would send the panels out, which would only work economically in larger batches.

The rear panel was nearly error-free (my measurements) luckily. It is hard to tell from the photo but the hole for the voltage select button needs to be moved up slightly to be centered. The hole for the RCA jack mounting screw could also be 0.2mm bigger. Everything else was right on the money and just fit right together.

The front panel needs a little more work. I had completely left off the hole for the 3.5mm input jack. :p Oops. I hand drilled the one in the photo below. The 1/4 inch and 3.5mm jacks need to be moved up about 0.2mm also, and the 3.5mm jack holes should be a little larger, 6.5mm instead of 6. The pot hole also needed to be 1mm larger.

So I'm going to send off another run of the panels with the adjustments and lettering this time and see how it comes out.


Here is something unrelated but interesting. In the solid state forum right now there is a posting in one of the threads that links to this slide presentation by the chairman and co-founder of Audio Precision on designing for ultra low distortion:

By Bruce Hofer, Chairman & Co-Founder, Audio Precision - Hosted by Bonis Audio - 2013 June

Skip to slide 18

Recommendations for Audio Circuits

and he concludes that 1206 sized thin film resistors are the best overall SMDs for audio use - exactly what I've used in the ODA signal path here.:) My main reason for using the 1206 size was to make it easy to solder, but it was in the back of my mind that the thermal change would be less with the bigger resistor (vs 0805 or 0603). RocketScientist / NwAvGuy wrote once that small SMD can actually be worse than through hole for that reason.

The presentation also suggests on that slide that very low temperature coefficient 1206 thin films are worth the money. Looking through what Mouser has it appears that the values of 5 ppm/C and 10 ppm/C he mentions are available, but largely special order and minimum 1000 quantity - and very pricey. I'm starting to see why Audio Precision analyzers cost so much. It looks like the 4.99K ground return resistor is avaliable in low tempco from Mouser at around $3.50 each. I'll add that to the BOM as an option along with the standard 25 ppm/C.

His second point on that slide under the 1206 size is one reason I stayed with larger through-hole resistors for the attenuation resistor and 1.5K feedback resistor, along with good current noise numbers. Since the gain stage can swing up to 14V peak on the output, I wanted to make sure the power dissipated was not just within the resistor's rating, but well within the rating to limit overall thermal effects in the resistor.
 

Attachments

  • IMG_1981.JPG
    IMG_1981.JPG
    217 KB · Views: 419
  • IMG_1982.JPG
    IMG_1982.JPG
    113.6 KB · Views: 403
  • IMG_1983.JPG
    IMG_1983.JPG
    134.1 KB · Views: 170
  • ODA front panel.PNG
    ODA front panel.PNG
    57.3 KB · Views: 182
  • ODA rear panel.PNG
    ODA rear panel.PNG
    54.8 KB · Views: 131
Last edited: