Supercharging Chord Dave's internal amplifier
Dec 10, 2022 at 12:14 PM Thread Starter Post #1 of 108

Reactcore

1000+ Head-Fier
Joined
May 17, 2010
Posts
1,359
Likes
2,040
Location
The Netherlands
See also the making of my Choral housed Mscaler and optical dual data link Here

And placing Dave's original PSU in external box Here

Edit: important notice at this post end!

Hi all,

Welcome to yet another modification of my Chord Dave.
This time its about making a add-on to its power feed to the amplifier section.

I'm not writing a introduction about the Dave DAC as that has been well covered in its own thread Here .
Dave is one of the most transparent and revealing DAC/pre Amps on the market available.

Since its output signal is processed and noiseshaped in the digital domain it could be equipped with a very simple analog output stage made out of discrete components. One key benefit is that this maximizes transparency due to the absence of analog filtering. Many, including myself, have noted that any external headphone amplifier in the chain degrades its transparency and its better to use its internal one.

But reviewers and other owners of a Dave have pointed out that its headphone driving capabillity is staying behind of when using some of the best external amplifiers connected to the line outs, especially when using harder to drive headphones.

So why would this be? its not the output voltage which at 6v should be more than enough to drive the toughest loads.
But many have questioned its switchmode power supply (SMPS).

During last years there have come different solutions for upgrading Dave's internal power supply by an external one.
But i didnt found much wrong with the sound coming out of Dave using my (quite easy to drive) Senns HD800's.

Until i got my ears on a Farad stack of Linear supplies hooked up to a Dave and i could A-B listen to it next to my stock unit.

What i heard supprised me alot.. i heard more pronounced mid-bass notes as well on my speakers as on my headphones.
Hm why would this be, in my reasoning its not so much about RF rejection since the SMPS should be taking good care of that as it is natively equipped with extending noise filtering cause of their own switching activity.
A transformer is not isolating a supply from net noise and so filters must still be applied for a LPS.

In my theory it had to be the Farad's heavy buffering by the use of capacitors with low Equivalent Series Resistance (ESR)

Ok so what does this do?
Imagine a voltage of 10v is supplied to a amplifier which is amplifying a sound signal.
Cause of this the amp draws a varying current out of the supply depending on the load it drives.
The power supply must keep its voltage steady but is showing a tiny variation of milivolts cause of the fact they exhibit internal resistances creating their voltage.

By nature PSU's are equipped with capacitors to smooth out the output voltage (ripple) which is either created by 10khz+ pulses (SMPS) or a double rectified sinus of 50 or 60Hz (LPS)

SMPS vs LPS ripple

Ripple.jpg

These capacitors also help to keep the total voltage steady when current is drawed from it.
But they have internal resistance in series with its 'energy container' over which their loaded voltage is devided.

ESR principle
ESR pricipe.jpg

Lower ESR means less voltage loss to keep the capacitor's charge on its leads steady.
Ok so due of lower ESR it can help to reduce voltage modulation cause by current drawn.
Bigger capacitors have a lower ESR compared to smaller ones, a capacitor's charge 'container' size is as many know put in farads.

Imo the importance is to get as low as possible ESR. This because the'container' wont get time to discharge as the frequency of the audio signal modulation is simply to high.. so a super high capacity (in F) is not most important but a internal resistor reacts directly by dropping the capacitor's voltage alowing the PSU voltage to drop.

Ok now i could create a external PSU loaded with capacitors in parallel to lower the combined ESR but then resistance is added again by the cabling, connectors and Daves PCB tracks before the power reaches Dave's internal regulator.

No i rather not allow a modulated voltage to keep existing over these internals, i want to add my buffercaps where its needed most: as near as possible to the amp.

So i measured inside Dave and located the two LM307/317 regulators which make +and- 12v out of the PSU supplied +and- 15v.

commentaar.jpg


Under a schematic of where i intended to place my capacitors, in red the added part
Dave schematic.jpg


Ofcourse many big sized (in farad F) capacitors wouldnt fit inside Dave so i went searching for smaller ones but as small as possible ESR i can find with at least 15v operating voltage.

I found newly released types with ultra low ESR made by TDK which are normally used in the electric automotive industry where theyre used to supply high currents for car motor regulators.

I went for the Axial types to have shorter leads in my Dave:

B40600 TDK.jpg

I picked the ones that still fit which are the 1700uF ones, but i place 2 in parallel for each voltage.
They cost as little as €6,50 each.

Enough theory.. lets get my Dave.

20221205_193420.jpg

First i start by making a 2.5mm massive copper leadwire and bow this in shape so i can lineup the capacitors.
wire.jpg

Then i soldered the capitors, two with the plus side and two with the negative side on it.
I bowed the other side's leads already in a way it can be installed fitting on the regulators.
20221210_124459.jpg



Now putting them into Dave.
The ground lead i soldered directly on the headphone socket's ground, this minimises ground currents flowing through Dave's PCB which probably lower groundplane noise further.
All can be placed without needing to take the PCB out.

20221210_125807.jpg


The +15v lead i soldered directly on the regulator's input pin.
Also this picture shows the ground lead on the headphone socket.
20221210_125834.jpg


On the negative regulator it is done the same.
But here the input is on the middle pin which is short combined with its baseplate.
With carefully alighning it can be soldered quite sturdy.
20221210_125847.jpg



A big help to let solder attach well and not to short the regulator's pins is the use of separate flux.
I used this one:
20221210_130026.jpg


After soldering it in, i just heated the ground lead again to take the tention off the other leads.
If all is done right it can be measured the voltages on the capacitors:

meten.jpg


So now im ready to listen to my 'Supercharged' amplifier inside my Dave.

And boy this thing got muscles now! i'm in my second day listening and am asthonished by the authority this brings, specially on higher volumes it holds grip on my headphone's drivers giving stronger and better defined bass notes without losing anything of Dave's famous transparency. Its a real joy to listen to all my recordings once again.

I hope you all enjoyed reading my long dragging post.. and.. happy modding :)
This guy dives in his music collection now..

Important Edit:
Soldering the ground lead on the headphone socket's ground pin caused the contact for the relay not to close and preamp mode was not activated when pulling the headphone jack.
This because it bowes the thick lead too much.

So place the ground lead on the bolt island aside it.

First make a circle on the end.
20221213_151140.jpg

The bolt island is also PSU ground
20221213_151101.jpg


Place the lead under the bolt.
I also soldered it on the island
20221213_153659.jpg


Now the headphone socket can freely switch the pre/hp relay.
 
Last edited:
Dec 11, 2022 at 6:00 AM Post #2 of 108
Another interesting project. You certainly keep yourself entertained with projects!

When Sean Jacobs was developing his cap board to go inside the Dave he let me try different versions with same value caps but made by different manufacturers and so with different ESR values. The initial assumption was that the ones with the lowest ESR would be best but listening to the various different ones proved this was not the case. Out of interest the lowest ESR ones were initially very impressive but compared to others they were ultimately rather fatiguing.
 
Dec 11, 2022 at 6:21 AM Post #3 of 108
Another interesting project. You certainly keep yourself entertained with projects!

When Sean Jacobs was developing his cap board to go inside the Dave he let me try different versions with same value caps but made by different manufacturers and so with different ESR values. The initial assumption was that the ones with the lowest ESR would be best but listening to the various different ones proved this was not the case. Out of interest the lowest ESR ones were initially very impressive but compared to others they were ultimately rather fatiguing.

It also depends on where theyre implemented.

I listened the whole day yesterday and i was far from getting fatigue. I also played a little time at louder level which before i avoided cause Dave started to sound a bit shrill then.. But not anymore.

It gives full rounded sound extending into deepest sub bass which balances nicely with Dave's high end making it an even smoother listen
 
Dec 11, 2022 at 6:48 AM Post #4 of 108
It also depends on where theyre implemented.

I listened the whole day yesterday and i was far from getting fatigue. I also played a little time at louder level which before i avoided cause Dave started to sound a bit shrill then.. But not anymore.

It gives full rounded sound extending into deepest sub bass which balances nicely with Dave's high end making it an even smoother listen

I’m pleased that your mod is paying dividends.

I still would want to hear the differences between different caps and select the best ones. We (wrongly as it turned out) thought that the first caps we tried were amazing and couldn’t be bettered. It would be easy to solder a wire connection to the regs so that different caps could be soldered to that for testing purposes and than the final selection of caps would be soldered direct. Of course because of the time taken to solder in new options this is all best assessed with two Daves side by side which is how Sean selected his final choice for caps.

Anyway, the main point is that you are doing amazing work so all credit to you!
 
Dec 11, 2022 at 10:23 AM Post #5 of 108
You are doing some great work there!

I did replace the original capacitors (by ones with lower ESR and higher capacity) in my Dave some time ago. I can confirm your listening impressions.

If I remember correctly, the original (470µF) capacitors are not paralleled on the output side of the linear regulators (like your picture shows). Two of the four capacitors are stabilizing the input sides of the regulators (one for the -12V regulator, the other one for the +12V) and the remaining two the ouput sides.

Is there a reason for you connecting the additional capacitors to the input side of the regulators?
In theory, they should be more effective if they would directly stabilize the output side. Did you test both locations?

As @Triode User already stated, it would be interesting to test different capacitor types. I did also 'only' select the capacitors according to their technical specifications (ESR, capacity, voltage, dimensions).

Keep up your great work:)!
 
Dec 11, 2022 at 1:42 PM Post #6 of 108
You are doing some great work there!

I did replace the original capacitors (by ones with lower ESR and higher capacity) in my Dave some time ago. I can confirm your listening impressions.

If I remember correctly, the original (470µF) capacitors are not paralleled on the output side of the linear regulators (like your picture shows). Two of the four capacitors are stabilizing the input sides of the regulators (one for the -12V regulator, the other one for the +12V) and the remaining two the ouput sides.

Is there a reason for you connecting the additional capacitors to the input side of the regulators?
In theory, they should be more effective if they would directly stabilize the output side. Did you test both locations?

As @Triode User already stated, it would be interesting to test different capacitor types. I did also 'only' select the capacitors according to their technical specifications (ESR, capacity, voltage, dimensions).

Keep up your great work:)!
The 470uf are in parallel for sure.
Its this way to keep the same pcb track length to the OP transistors

I have tried them after the regulators first.. even directly on the transistors.. but it gave distortion probably caused by intermodulation with the other capacitors.
I havent tried to take those out though..

20221205_203130.jpg
 
Last edited:
Dec 11, 2022 at 4:50 PM Post #7 of 108
The 470uf are in parallel for sure.
Its this way to keep the same pcb track length to the OP transistors

Thanks for making this clear. Yes, this makes sense. So the inputs of the voltage regulators (+/- 15V) are only buffered by capacitors located far away from the regulators. Long PCB traces, wires and the Molex connector between the capacitors and the regulators...

So I'm sure, adding the capacitors the way you did - directly at the regulators - will result in a significant improvement!

Your mod does look very appealing to me! I guess there's no way around now for me:)...

I did just replace all original capacitors by better types: 1000µF/25V with an ESR of 17mOhms from Nichicon (Type UHV1E102MPD). They are drop in compatible with the originals. Their greater height doesn't matter in the Dave housing.

Thanks, Martin
 
Dec 11, 2022 at 5:11 PM Post #8 of 108
Thanks for making this clear. Yes, this makes sense. So the inputs of the voltage regulators (+/- 15V) are only buffered by capacitors located far away from the regulators. Long PCB traces, wires and the Molex connector between the capacitors and the regulators...

So I'm sure, adding the capacitors the way you did - directly at the regulators - will result in a significant improvement!

Your mod does look very appealing to me! I guess there's no way around now for me:)...

I did just replace all original capacitors by better types: 1000µF/25V with an ESR of 17mOhms from Nichicon (Type UHV1E102MPD). They are drop in compatible with the originals. Their greater height doesn't matter in the Dave housing.

Thanks, Martin
The molex is only included for the ones on the PSU.. aside it on the pcb are another three 1000uF with some filters (which also add resistance in the chain btw)

Both sides of the regulator should have rock steady voltage.. if the input fluctuates.. the regulator keeps compensating to keep the output fixed.. this can cause reverse modulation on the amp's power feed.

Let me know if youre making one too🙂
 
Dec 11, 2022 at 5:38 PM Post #9 of 108
This is an early prototype of the commercially available Sean Jacobs cap board. I think the final one only varies from this with regards to the wiring between the board and the molex.

What was interesting as part of the beta testing for the options for the Sean Jacobs cap board was that the 5v digital rail was at least as susceptible to the improvements as the analogue rails, if not more so. Again the actual caps used were very influential to the improvements. So it might (will) bear fruit to look at the 5v rail.

EC7246BB-56F6-4F81-AD9F-1BBE7EF9E524.jpeg
 
Dec 12, 2022 at 2:28 AM Post #10 of 108
This is an early prototype of the commercially available Sean Jacobs cap board. I think the final one only varies from this with regards to the wiring between the board and the molex.

What was interesting as part of the beta testing for the options for the Sean Jacobs cap board was that the 5v digital rail was at least as susceptible to the improvements as the analogue rails, if not more so. Again the actual caps used were very influential to the improvements. So it might (will) bear fruit to look at the 5v rail.

EC7246BB-56F6-4F81-AD9F-1BBE7EF9E524.jpeg
It might sounds odd
But i wonder how it performs if you would feed this board with the stock SMPS. Provided it can initially charge it and not go in protection mode.

Imho and experience its the buffer that counts.. not having half a Kw power overhead.

Thats why i chose not to make a external PSU (or my 5 and +/-15v battery design)

Then again.. i never heard a SJ
 
Dec 12, 2022 at 2:48 AM Post #11 of 108
It might sounds odd
But i wonder how it performs if you would feed this board with the stock SMPS. Provided it can initially charge it and not go in protection mode.

Imho and experience its the buffer that counts.. not having half a Kw power overhead.

Thats why i chose not to make a external PSU (or my 5 and +/-15v battery design)

Then again.. i never heard a SJ

An early experiment I did before getting the SJ power supply was to put the stock smps in an external case. Getting it out of the Dave seemed to be better , putting ferrites on the umbilical helped further but neither helped as much as the SJ.

Also, when I was going for the latest version of the SJ I wondered if I could get away with not upgrading the 5V supply, thinking that the analogue rails were probably more critical. That turned out to be a forlorn hope because the upgraded supply to the 5V rail made as much of a difference as it did to the +/- 15V.

In answer to your question about whether the local caps within the Dave mitigate the advantages of SJ over the factory SMPS, I have not tried that but I did try a similar experiment between the earlier SJ DC4 and the later SJ ARC6 where both were feeding the cap board. That showed that even with the cap board the later ARC6 was still better.
 
Dec 12, 2022 at 5:39 AM Post #12 of 108
An early experiment I did before getting the SJ power supply was to put the stock smps in an external case. Getting it out of the Dave seemed to be better , putting ferrites on the umbilical helped further but neither helped as much as the SJ.

Also, when I was going for the latest version of the SJ I wondered if I could get away with not upgrading the 5V supply, thinking that the analogue rails were probably more critical. That turned out to be a forlorn hope because the upgraded supply to the 5V rail made as much of a difference as it did to the +/- 15V.

In answer to your question about whether the local caps within the Dave mitigate the advantages of SJ over the factory SMPS, I have not tried that but I did try a similar experiment between the earlier SJ DC4 and the later SJ ARC6 where both were feeding the cap board. That showed that even with the cap board the later ARC6 was still better.
In your early smps external experiment you didnt have the cap board yet i guess.

Theres also the RF subject influencing this..
In my situation this is cared for by using my optic link.. so my case differs a bit..

I could focus on the current reserve with this mod.
 
Dec 17, 2022 at 7:09 AM Post #14 of 108
Super nice. Thanks for the detailed modding steps. Wish you lived near me. Would love you to mod my Dave.
 
Dec 17, 2022 at 7:19 AM Post #15 of 108
Super nice. Thanks for the detailed modding steps. Wish you lived near me. Would love you to mod my Dave.

Its a easy mod actually.. im sure if you find someone with soldering skills you can let it be installed. The caps can be bought at Digikey.. they ship worldwide to consumers.

I see you have the Farad3 i wonder how they combine as im on stock PSU.
 

Users who are viewing this thread

Back
Top