Thanks. Whew! Hence the acronym.
Excellent info, thanks
Very informative that the Aps Lessp is inline with the PS Audio 512 flagship.
I wonder what the designers at PS Audio would have to say, after all that work and money spent and the Aps Lessp which is a modded PS Mk1 is close or better in league to their flagship 512.
PS could easily implemented these changes instead of reinventing the wheel.
I have to get an Aps Lessp to find out, since I am true devotee of 2008 Mk1, dsd modded.
We canāt assume that one personās opinion decides that the modded DSD Mk1 matches or surpasses the 512.
Or even a group of 4 or 5 audiophile buddies.
FWIW, my mk1 is medium-level modded with edcors and the external LPS, with an Auralic G1 into a Denafrips Hermes feeding the mk1 i2s. I also have the PMG and compared using itās internal streamer, and yeah the PMG was clearly better with zero break in right out of the box, so Iād expect the gap to widen with break in.
However, for 1/3 of the price, you canāt argue with the value of the Hermes + modded mk1.
The rest of my system is a Gryphon Diablo 300 and Revel Studio V1s with iconoclast and furutech cables.
Allears has said in another thread that the LESSP is a significant sound improvement from the original mod that you have.
Hardware upgrades are more expensive to implement than software. Iām guessing that the hardware parts in the APS Lessp cost more than the parts in the PMG. But the PMG Iām sure has more advanced software and computing power.
APS published the DC offset correction. And I finally understand what they meant when said: swapping R5-R6 could worsen the situation.
They were right! I was wrong. The problem is real.
If someone with stock 1% resistors measured low differential DC offset [like mine 0.03mV and 0.14mV], then swapping with higher tolerance resistors will most likely raise the DC offset to around 0.3mV. Because of the internal DC offset of the opamps used.
Hereās why. Ted Smith explained:
On the Mk I, the input common mode to the opamps is 3.7V. The midpoint of the power supply is 3.5V. That 0.2V difference is a part of the source of the DC voltage on the inputs to the transformer. Change the R5:R6 ratio to from 1:1 (= 3.5:3.5) to 3.3:3.7 to raise the input common mode voltage of the opamps to 3.7V. That will lessen the voltage difference on the transformer inputs.
It means we can use common values: R5 = 475 Ī©, R6 = 523 Ī©. This gives: Vocmā3.67V.
When APS use their method with fixture, theyāre compensating all the resistors deviations (and much more). This is probably the most precise method. But (there is always But) without swaping resistors R1 through R6 with better ones (why R1-R6 - explained in my previous post about DC offset correction) with warmup of the DAC the DC offset might drift! Because of the Temperature Coefficient of the stock Thick Film resistors. Iām not sure which stock resistors were used, but typical 0805 1% resistors could have this coefficient as bad as Ā±100 to Ā±200 ppm/Ā°C.
It depends on how hot the DAC gets. We know that this DAC could get pretty hot. As I mentioned before, donāt stack it with other gear.
.
I think we could do something without a test fixture.
-
use R5 = 475 Ī©, R6 = 523 Ī©
-
use 0.1% thin film resistors (in my calculations, I showed that itās more than enough) for all R1s through R6s.
-
use 1206 size! - will run cooler due to its larger surface area. Less self-heating and thus a more stable actual resistance
1206 Panasonic ERA-8A has Temperature Coefficient Ā±15ppm/Ā°C (in some values - Ā±25ppm/Ā°C).
.
Iām super excited to try itā¦ I will put it on my long to-do list.. maybe someone else will try it sooner.
BTW, MkII has pretty much the same DC offset situation. With Edcore 4400 this doesnāt hurt, but with APS transformers it might.
Iām not sure if it was discussed there. Iāll ask, maybe they want to fix their DC offset, too.
According to APS, The MK2 doesnāt have the same Vocm as the MK1. It uses an active servo circuit.
From APS:
āOur R2 transformers experience less impact from DC offset compared to the factory transformers. The MK2 DACs possess significantly lower if any DC offset (~0.3mv compared to the MK1 DACs since the MK2 utilizes an active servo circuit to divide the supply voltage. DC offset is not an issue with the MK2.ā
According to the designer of the MkII, Ted Smith, thereās no active servo:
The input common mode for the opamps is 2VDC and the output common mode is Ā½ supply which is 5V so the opamps see a -3V common mode input overall. The opamps have a 80dB DC CMRR which is a factor of 10,000 so thereās an expected 0.3mV of DC on the transformer inputs. The stock transformers arenāt affected by this DC offset, if anything it shifts the hysteresis curves a little giving a more linear response from the transformers, sort of like a class AB amp.
The Mk I uses AD8132 video opamps which can have a CMRR which in the worst case can be 10 times worse. But probably not that bad. On the other hand the input DC for the Mk I is about 0.2 V instead of 3V so that just about balances.
As far as I understand, the opamps in MkII are better (CMRR is 10 times better), but the input DC is 15 times higher than in MkI. Which means the situations are the same for the resulting differential DC offset.
New info from APS website.
Our new type R3 transformers experience less impact from DC offset compared to the factory transformers. The MK2 DACs possess a significantly lower DC offset (~0.3mv compared to the MK1 DACs, since the MK2 utilizes an active servo (regulated) circuit to divide the supply voltage. There is a separate regulated analog 10v supply and a regulated analog 5v supply rail to supply the opamps Vcom. NO RESISTOR DIVIDER is used. DC offset is not an issue with the MK2.
