Moving from 9038Q2M to a 9028Pro board

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The 9028PRO board with the 'golden clock' you can buy on ebay for about $10 (only the clock, not the board) is not a low phase noise clock so far as anyone has been able to show.

The IV stage doesn't use quality thin film resistors and C0G caps in the signal path.

The voltage regulators, some of them anyway, probably need to upgraded.

The AVCC supply is one regulator for both left and right channels, so stereo separation is probably affected as would be expected.

It still does't have any way to upsample, either. My current recommendation for that may be changing from an SRC492 to an AK4137. I made some comments as to why in another thread, but I can post a link here if anyone is interested. I will have more to say about it later anyway after some more tests. The basic rationale for upsampling, which is the reason Benchmark and Crane Song use it, is to fix reconstruction filter issues that are a part of all hi-res DAC chips currently made (according to Benchmark).

Bottom line, I would agree it appears the 9028PRO board is better as-built than the 9038Q2M board. I don't know how much work or if it would even be possible to make the 9028PRO board sound as good as my modded 9038Q2M board. That remains to be seen. If the AVCC issue can be fixed then it can probably be made as good as my modded board, but the end cost will be more because there is still a lot to fix to make it the best sounding DAC it can be.

As far as the question of how good is good enough, I would say that I have been encouraging people to mod a headphone amp to go with the DAC mods. An high quality HPA would be pretty cheap to do, a whole lot less expensive than a power amp, and it will let you know how good your DAC actually sounds. Getting back to how good is good enough, it may depend on how good you have heard. Too much listening to a really good system risks making you want one. The safest protection is never listen to a good system, then you won't know how good it sounds and you can remain satisfied with whatever you have right now. For myself, I guess I can listen to a some car stereos, but not all, and I can listen to iPhone audio some, but none is good enough for a main stereo. Since I know what a good one sounds like (potentially habit forming), that's what I want to have. I also want it to be able to serve some utility function such as be usable to mix a record with if the need arises, which it does now and then when people track me down asking me to do it for them.

Lest we forget that we need to put all of this into a case at the end. That was when it hit me that some kind of compromise will be needed because fitting a mish mash of a ton of boards into a case is not easy and the amt of metalwork is huge. When I saw that a decent case was available that fit this exact dac board and I estimated just enough space to fit everything in, including possible mods, that sealed the deal. I hate doing intricate metalwork, and when you have a display, it needs to "look" right. PLUS it has a remote.

Yeah, the upsampling, reconstruction and THD compensation etc. will be missed but I think even the Buffalo series does not have this either. I guess there are three levels, things like the 9038Q2M, boards like these, then things like what Twisted Pear is doing. I was intentionally trying to avoid too much serious modding on the replacement DAC board. Remember that caused the smoke to escape on my last DAC.

And in the end I keep reminding myself that this DAC will cost less than the taxes I would have to pay on a DAC3 purchase!
 
IIRC, I think the Buffalo DACs have an SRC4392 for SPDIF inputs only. Don't know if they use it to upsample at all.

But, the main thing is I wouldn't necessarily do something or not just because of what someone else does. The only thing I would really encourage people here to do would be to mod a HPA, invest in some accurate headphones and use those things as test instruments and for enjoyment. Once you can hear what you are doing then you can decide if your DAC is good enough or not. If you can't even hear what you have, how you can make a decision about what to do next?

Also, because of the expense of SRC and because most people aren't familiar with what they do I have offered to loan one out to somebody who has done the other mods and wants to hear what the SRC can do for the sound as a final mod. In fact, I think I might offer to send you one to try out since I have a spare that can serve as a loaner and your 9028PRO DAC already sounds not-too-bad as-is. Would you be interested?
 
The $140 ES9028PRO board seems to have most of the quality features we would like to upgrade to. The 3 totally separated PS, I/V output and a better clock. I estimate adding USB input, PS source and a pre-drilled case will add up to about $300. The sound would have been good enough for most of us without further board upgrade.

The SU-5 assembled ES9038Q2M DAC has I/V and ready to run at $200. A good quality linear PS box will add about $30. I am waiting for Markw4's assessment on how good it sounds as is and quality of the parts, including clock used.

I have an assembled ES9038Q2M DAC with TFT display, USB for $120. I am about done upgrading it to LDO dual linear PS. But it has no I/V and the clock is not so good. It sound pretty good to me and like the filter adjustment, but I do not have a reference for high quality DAC. :confused:

Two low end, commercial ES9038Q2M DAC, Pro-Ject Pre Box S2 Digital and Topping D50, both get very good reviews without the I/V. But they get good separate PS. I am facing the question of how good is good enough. I am very tempted to get the SU-5 DAC if it sounds better than the ES9038Q2M board that started the ES9038Q2M thread. Any first hand input on comparison of the above mentioned 3 options is appreciated.


Hello

If had any advice to give :rolleyes:

Change the power supply step by step to see the impact on the sound .
In fact I changed the 3 at the same time and I do not know if the 3 are
really necessary!:D

Serge
 
Lest we forget that we need to put all of this into a case at the end. That was when it hit me that some kind of compromise will be needed because fitting a mish mash of a ton of boards into a case is not easy and the amt of metalwork is huge. When I saw that a decent case was available that fit this exact dac board and I estimated just enough space to fit everything in, including possible mods, that sealed the deal. I hate doing intricate metalwork, and when you have a display, it needs to "look" right. PLUS it has a remote.
Well said.

When I first start the DAC project, I purchased this enclosure that was specifically pre-drill for the VR1.06/1.07 ES9038Q2M board.
MINI aluminum chassis case for ES9038Q2M DAC DSD Decoder | eBay
I still have it on my shelf. (PM me if you are interested) But it does not have the cutout for the display. So I ended up purchased the fully assembled DAC with the Q2M board, USB and display.
 
Last edited:
Update. I did install the Sulzer Borbeley Power Supply first.
I will show step by step details.
Turned it on and discovered something. Uh Huh... I was always told to watch out for this. Fixed it.
Then I listened and listened and the conclusion will come later.
Then I went for it and decided to install a couple of LT3042 boosted to 1 amp.
I modified each LT3042 board with a Cset of 47uF Tantalums + whatever was already on the board. I then crossed my fingers that the slower ramp up would not cause issues.
None. Turn on was smooth as silk.
Listened and listened. Wow....this time the LT3042s sounded great in the low end unlike my experience I had with them on the 9038Q2M. There it probably was not the LT3042 but something else.
Bass is fantastic. Chorals...superb, Brass with real bite. Pianos will pound your chest when the player wants. But when you want soft vocals, the sound could indeed seduce. Yah.
I'm a fan of PRAT and this has it all in spades let me tell you.

Advance thanks for MarkW4 whose work on the 9038Q2M board inspired me to get into better digital and I am glad I did follow his guidance. Learning that smoke indeed runs ICs. I am keeping it inside this one!

Over the next few days I'll post pictures and outline everything and my impressions. What the analog supply effect is. What the AVCC effect is.
But not to keep you waiting....this board, IMO is a much better base to start from despite it having some compromises. I will outline why I feel so and the decision which direction to go for those sitting on the fence will be up to you but you need to know the pros and cons.

This weekend is going to force me to upgrade my crossover network for my subwoofers and I will tell you what I found there as well.
 
Last edited:
While I had the board out I would show the undersides.
The lighting was adjusted as best I could to enhance the line shadows.
 

Attachments

  • IMG_1088.JPG
    IMG_1088.JPG
    957.5 KB · Views: 389
  • IMG_1089.JPG
    IMG_1089.JPG
    929.3 KB · Views: 389
  • IMG_1090.JPG
    IMG_1090.JPG
    662.2 KB · Views: 375
  • IMG_1091.JPG
    IMG_1091.JPG
    942.9 KB · Views: 365
Now we begin the mods.
First we remove the 317/337 regs.
Since the regs were removed, there was no need for the voltage dividing resistors to the adjustment pin so I elected to remove those as well as the decoupling cap.
Then a thought occurred to me...... I could put in some pilot lights in there by reusing the 2.4K resistors and some LEDS. That way I can easily tell when my regulator has output.
Next stroke of luck was that there was a large plated hole for the ground point. I grabbed some 12 gauge wire and used that for joining ground to my Sulzer Borbeley PS.
 

Attachments

  • IMG_1092.JPG
    IMG_1092.JPG
    915.7 KB · Views: 371
  • IMG_1093.JPG
    IMG_1093.JPG
    881.6 KB · Views: 150
  • IMG_1096.JPG
    IMG_1096.JPG
    884.7 KB · Views: 144
  • IMG_1095.JPG
    IMG_1095.JPG
    974.8 KB · Views: 134
  • IMG_1097.JPG
    IMG_1097.JPG
    959.6 KB · Views: 151
  • IMG_1098.JPG
    IMG_1098.JPG
    867.1 KB · Views: 149
  • IMG_1099.jpg
    IMG_1099.jpg
    220.7 KB · Views: 151
  • IMG_1101.JPG
    IMG_1101.JPG
    957.1 KB · Views: 148
  • IMG_1103.JPG
    IMG_1103.JPG
    878.1 KB · Views: 148
Now comes the shutting down of the LT1963s.
One LT1963 feeds the AVCC lines, the other LT1963 feeds the Clock and other voltage requirements of the DAC via two AMS 1117s.
Since space was going to have to be minimized to fit into the custom case I also purchased from Ebay to fit this DAC. I started to think about compacting everything.
I believe in preregulation rather than brute force capacitance. So the Ebay LT3042 modules I had to needed to be reduced in footprint since space was going to be an issue. I'll let you know later why I did what I did and what it means down the road.
So I removed the large Capacitors on the LT3042 module and installed some more compact input caps and also removed the rectifying diodes as that work will be carried out on a 317 preregulator. ( you could reuse the 317 removed in the first step to fab a preregulator or do what I did purchase an inexpensive 317 single reg board from ebay for about $2...upgrade caps though.)
I then modified the 1A Current boosted LT3042 module to 3.3V and then installed a 47uF tantalum capacitor as Cset. Why 47Uf and not 22uF. This was left over when I was trying to build the op amp AVCC and my 9038q2m was fried in the process.

I also added an LED to the board again to show power to the board was good.
You'll notice I am using through hole stuff to mount on SMD intended pads. It's all good but you need to take your time and I also used all kinds of contraptions to hold everything in place before soldering. This was nearly as complicated as created making my custom 744-811 module on an 8 pin DIP over 20 years ago.

In the end all good.
 

Attachments

  • IMG_1105.JPG
    IMG_1105.JPG
    875.7 KB · Views: 189
  • IMG_1106.JPG
    IMG_1106.JPG
    902.7 KB · Views: 194
Last edited:
Can we provide separate feeds for AVCC L and AVCC R. Looking at the board closely I do believe it can be done. It would involve cutting some traces and making wire runs directly to the L and R reference voltages and to two decoupling caps for AVCC.

Will I do this? I'll think about it but the limitation I had stated up front as regards AVCC lines is likely not there anymore but will depend on how much modding one wants to get into. As it is now, I can easily remove all the mods and return the board to stock. So there was no invasive modding at this point.

What more could have been done? Of course Clock changes......could be in the works.
Another LT3042 separate voltage feed for the clock...another $15 and SPACE and crystal......Never changed crystals before.

Upgrade decoupling caps.

Install shielded wire runs from the LT3042 feeds to the DAC board. Same for Analog. PS. Remember only one side of shield to be connected

Finally, I am thinking of removing the LT1963s from the board and use the rectifying circuit on the board. Its output will feed the LT3042s but each LT3042 board will have its own preregulator. This WILL then allow enough space in the case to fit possibly 4 - LT3042 modules into the case.....thus AVCC L. AVCC R, Clock, and feed to AMS1117s.
This would require invasive modding....remember I did not have to fabricate any boards myself.

I could also upgrade the resistors on the board to higher quality metal films.

At this time the sound is excellent in my opinion and I need to look at something else.

For those thinking of this DAC, while I did not fabricate any boards the analog power supply board is also very important. I had some left over from the early 80s and to this day they still sound excellent. These can be bested like I did to my preamp by Jung Super Regs or even his newer Shunt regs using similar design. ( Other low Z low noise regs can also be used but I can't comment on their performance) I have been a keen follower of Walter Jung since the early 80s and really appreciate his contributions to the audio world. Boards for the Super Reg can be purchased from the DIY Audio store and will indeed be better than the SULZER Borbeley. How much better will depend on how low your system can go and how much air your system can move. I am using VMPS Larger subs driven by Leach Amps with fully regulated power supplies and they do not poop out. The Super Regs add a sense of space that only the lower octaves will provide. If your system does not dig that low, I am not sure there will be a difference between the Super Reg and the Sulzer Borbeley.
I am now really thinking about going to the DIY Audio store.
 
Last edited:
Mike, Thanks for all the pics and descriptions. Glad to hear you are happy with how things are coming along. From my knowledge of Sabre dacs I would say you aren't all the way to where you really could be yet, but I will have some more later that may be of interest. I am expecting an ES9038Q2M board that looks like it already has IV output. If it does, I will let you know how it compares with the fully ESS recommendation compliant modded board I also have.

Also, please be aware that the LT3045 data sheet says that going with more than 22uf for Cset will reduce the regulation bandwidth of the LT3045. Presumably, LT3042 would be the same but I haven't checked.

Lastly, I still have an SRC4392 board that is burning a hole in my pocket. I am looking for a volunteer to try it. Somebody will have to step up to the plate and give it a go. How about you?
 
Low Noise LT3042 Linear Regulator Power Supply Board DC Converter Overvoltage | eBay

I just purchased a 47uF 25V bead type tantalum off the shelf at my local parts store.

Then I removed the 47K SMT resistor and soldered on a 33K metal film resistor in its place. If you can do SMD and have the equipment for it then do so.
The tantalum cap solders on the R1 pad as R2 and R1 are pads are paralleled. Same process of soldering.

Simple AVCC power supply. It is much much better than the LT1963. Transforms the DAC and unlike my prior experience, no low end issues anymore. Better than 15uF of films as Cset.

The fact I do not need to make a board is whole big advantage. I think when picking an LT3042/5 make sure you have the facility to add a larger Cset. This was brought up by MarkW4 and should not be ignored.

I see some are using LT3042 to power op amps. I don't know if that is better than a op amp based lowZ low noise reg.
 
Mike, Thanks for all the pics and descriptions. Glad to hear you are happy with how things are coming along. From my knowledge of Sabre dacs I would say you aren't all the way to where you really could be yet, but I will have some more later that may be of interest. I am expecting an ES9038Q2M board that looks like it already has IV output. If it does, I will let you know how it compares with the fully ESS recommendation compliant modded board I also have.

Also, please be aware that the LT3045 data sheet says that going with more than 22uf for Cset will reduce the regulation bandwidth of the LT3045. Presumably, LT3042 would be the same but I haven't checked.

Lastly, I still have an SRC4392 board that is burning a hole in my pocket. I am looking for a volunteer to try it. Somebody will have to step up to the plate and give it a go. How about you?

Check your PM about the SR converter.
 
Ok, sound impressions.
Moving from 5532 to LME 49720 and LME 49710. Totally stock DAC board.
The 5532 has a thicker sound than the LMEs. Initially it may sound more powerful because the tilt is such that the midbass is more prominent and there is definitely less high frequency energy. You get a hint of the higher frequencies. If you have a hot metal dome tweeter, this might be able to tame it for you. The midrange is certainly veiled.
It is pleasant sounding and should not cause anyone to run away when hearing the sound. I will give it a 7 out of 10.

The LME497xx on the other hand sound less powerful. There is a lot less veiling and the midrange is clearer but not in your face. The high frequencies are definitely more distinct than the 5532/5534. Because there is more energy at the upper end, the low end feels light. Listening carefully it is all there though. Just not distinct enough. The 5532 sounds stronger there but it is not clearly defined. Some level of muddiness exists in both families of chips but there is less of it in the LME 497xx.
Remember this is with the stock DAC using LM317/337 regs. This is an important distinction as you will come to see.

I still await the AD797s to replace the LME49710s. From an imaging standpoint on the stock DAC at this stage. Nothing really different between the two, shallow depth of field, fuzzy imaging. Can't say it was good.

Next installment, after the 317/337 three term regs were replaced. That is when it gets interesting.
 
You might try adding some more proper bypass caps to those LME49720 family opamps. they can go to higher frequencies than a few film caps can deal with. The decoupling should be right at the opamp pins and to the ground plane.

Then, to clear up the rest of the problems you attribute to the opamps you will need some good thin film resistors and C0G caps for the signal path. That and an ultra-low-jitter clock will get you closer and help clean up most of the funny stuff you still hear.

The final bit will have to come from really good upsampling, but you should be able to hear improvements as you go through each remaining problem area.

Reason the bass still sounds a bit thin is because there is some remaining distortion. It brightens up the bass even though the low bass is fully there.
 
Last edited:
The distortion if it was that was completely cleared up with the addition of the Sulzer Borbeley power supply. The compression at the low end was now gone, extension was increased, midrange opened up a lot more, the top end was free of any excesses. The thinness of the sound was removed. This was before any mods to the AVCC.

What did not improve that much was the sensation of space where you could easily perceive the space the recording took place in.

I've heard all the arguments about PSRR of op amps and that the need for a high performance power supply is not required. This experiment proved once again that for whatever reason, the need for high performance power supplies is needed more acutely the higher performance the IC becomes. Provided you have a system that resolves it and listen carefully. ..well in this case, you actually did not have to listen carefully. It was so evident.

So at this point, I will state that a 317/337 reg simply does not cut it for power to the IV stage. You need something good, how good? Heck my Sulzer Borbeley i pulled out of my box dates back to 1983! There are even better ones today, the reason I pulled these was that back in 2000, I put in Jung SuperRegs and the improvement in bass extension and dynamics out of my preamp was so evident even with that old DAC the Sonic Frontiers DAC2. Vinyl not so much. So if you see a DAC using a 317/337 setup you need to pull those out. Again there are newer power supplies that might be as good or better but a 3 term reg....are only suitable as pre regs these days.

What does that mean, even with a proper IV circuit, you really need a proper high performance power supply for it.

At this point, I pondered about what Sergelisses had said that the LT1963 might be good enough. That did a couple things....made me wonder whether or not it could get better or whether the LT1963 was indeed good enough. The improvement was so dramatic that I really did think that maybe the LT1963 would stay in and that the LT3042 would sound the same and save me a ton of work.
Had lunch and then came back and said What the heck. I was itching to find out. Yeah, I did not have the 22uF tantalums and I had 10 or 47. 47 it was.

Started modding the LT3042 modules.

I could have removed the LT1963s off the board but I wanted a way to reverse course if the LT3042 did not work out. So as previously planned I was going to tack on the LT3042 circuit. Tacking on was easy because the tab on the LT1963 is Ground and the two rightmost pins can be used to tack onto. A fairly large contact patch.

If it didn't work out, I could easily desolder and go back to stock AVCC.
 
Last edited:
Okay then. Still, however you cut it people find that better clocks help too. As do the best passive components. It may seem like nothing could get better than it is right now, but then you don't have a DAC-3 there to keep you calibrated.

It can be like, "Wow!!! This DAC sounds great!!" Then you go over to check with the DAC-3, and you say, "Hmmm... looks like I have some more work to do..."
(of course when I say 'you' above, you know who I mean.)
 
Last edited:
Okay then. Still, however you cut it people find that better clocks help too. As do the best passive components. It may seem like nothing could get better than it is right now, but then you don't have a DAC-3 there to keep you calibrated.

It can be like, "Wow!!! This DAC sounds great!!" Then you go over to check with the DAC-3, and you say, "Hmmm... looks like I have some more work to do..."
(of course when I say 'you' above, you know who I mean.)

maybe mine sounds better than the DAC3 especially when I have a sip of that drink and a bunch of coins jangling in my pocket!!!!!!!!:D:D:D:D
 
But seriously at some point, euphonics does play a part. Up to a certain point you can go and objective, then it starts to become system specific and euphoncally. That is life.

Anyways I forgot to mention after I changed the power supply, darn the right channel sounded distorted but I had not messed with the circuit. OMG moment...not again.
Then I switched cables and indeed the source was the DAC. The only thing.....those darn ICs in those sockets. I reseated the right channel ones. Fixed. I used to be told since the 80s, always solder in the ICs for best sound. I know why now.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.