HOLO Audio MAY DAC

[GoldenOne]

Just now, k27R said:

Does anyone know if the RCA outputs and XLR outputs be used simultaneously without degradation in quality?

They can be used simultaneously but I think you'll lose CMNR on the balanced because one 'half' will have an impedance mismatch to the other. 

 

This is the same with basically any balanced dac though unless it specifically has a separate output stage for SE

[GoldenOne]

11 hours ago, Extreme_Boky said:

You can't "lose" CMRR... the amplifier section will benefit if the analogue signal is transmitted as balanced, from the balanced source (DAC), due to the capability of the differential input stage (located in the amp) to reject the common noise. In other words, the overall sound will have fewer distortions - especially 2nd harmonic... and further even harmonic distortions, as well as less noise), if the balanced transmission is used.

 

If only one half of the balanced signal is used, then that half is referenced to ground and it must have ground/shield connection as its return path - where balanced actually can work without the shield, but this is not recommended.

 

I believe that May have separate (isolated from each other) balanced and single-ended analogue stages. One of these days I'll have a look inside to be certain.

 

On this note, I prefer to use half of the balanced output as a single-ended output; this approach does not require an additional OP amp (or discrete transistor circuit that may require 4-5 transitions) for balanced-to-single ended conversion.  This is strictly when only one output (either balanced, or single-ended) is required at any one time.

I was meaning CMNR on the interconnect not the output stage. external influences could be a concern

[GoldenOne]

10 minutes ago, BrownMagic said:

Will a standard I2S cable work between the May and Pi2AES? 

Yes. Just set the May's I2S pinout to the "ALT1" layout and it matches pi2aes.
The configurable i2s pinout is a godsend

 

10 minutes ago, BrownMagic said:

Also you mentioned about locking. If PLL is turned on, does it have to keep locking each time the song’s bit rate changes?

It must lock every time the sample rate changes. Bit depth isn't an issue. But most music is gonna be 44.1khz anyway so you won't have an issue often.

 

11 minutes ago, BrownMagic said:

Is there any scenario when turning off the PLL will be advantageous to SQ? 

As I understand it the only time this would be beneficial is if the local clock is poorer than the source clock. But, given the incredible performance of the May's internal clock system, not just in terms of the clocks themselves but the actual implementation, this is VERY unlikely to happen. I've not found any digital source with outperforms the may's internal clocking thus far so It's best to leave the PLL on.

 

 

13 minutes ago, BrownMagic said:

Does PI2AES benefit from PLL being on?

Yes, there is a slight improvement in jitter performance with the PLL on. Though the pi2aes jitter performance with the PLL off is still exceptional.

 

[GoldenOne]

I've put part 1 of my May review up :)

 

[GoldenOne]

Part 2 now up

 

[GoldenOne]

Just now, ThenewGearPPK said:

Does anyone have any problems getting PCM 1.536MHZ to run on May KTE with HQPlayer via direct USB?

 

Every time I select 1.536, the May just detects 705.6k instead & won't go higher.

Even jRiver can get it to recognize up to 786k, but not on HQplayer.

Is this a bug or is this fixable? A PC problem, USB port problem, driver problem?

 

The USB driver info shows version 5.00 & the May as revision v30.14.

 

Its cause you're on 30.14

 

30.12 supports 1.536mhz but may not work with non-Intel usb chipsets. 

 

30.14 is the stock firmware because it works with any USB source, though only goes to 768khz

[GoldenOne]

11 hours ago, Extreme_Boky said:

That May in the review is running v1.30; mine is running v1.20. Does any one know what the difference is?

 

1.2/1.3 is the hardware version not the USB firmware

 

 

11 hours ago, ThenewGearPPK said:

Thanks, I am on a AMD chipset so that would be it.

 

BTW how would I downgrade to 30.12?

Do I just run the holousbaudioDfu.exe file in the driver folder or do I have to uninstall something first?

30.12 isn't a downgrade it's technically an upgrade. 30.12 and 30.14 were released at the same time they're just different.

If you're on AMD then you'll need to stick with 30.14 and won't be able to use 1.536mhz.
30.12 caused problems with my AMD pc so I'd advise not trying to swap to it.

I'm using one of these as a roon and HQP NAA endpoint currently: https://www.amazon.co.uk/MeLE-J3355-PCG02-APL-64GB/dp/B08T6CT5VL
Not too pricey, intel, gigabit LAN, silent/fanless too. Works great. That will allow you to properly use 30.12 and 1.536mhz

 

 

8 hours ago, Roasty said:

Same thing I experienced with my amd pc. Its just not playing well with 30.12. Just use 30.14 and be content with 768 until Holo comes up with a fix. 

Unfortunately it's not something holo themselves can fix. The issue is apparently with the source chipset not the DAC.
I couldn't get 1.536mhz working on denafrips on my AMD pc either but was fine on intel.

There are some non-intel machines which work but it's very hit or miss.

[GoldenOne]

8 hours ago, Quadman said:

Maybe v1.30 is the reason why people are reporting no sound output with firmware 30.12 with AMD based PC and with 1.20 sound is output, maybe not 1.536 with amd but 768.  I have not seen any published description of the changelog from 1.2 to 1.3

I asked Jeff, 1.20 fixed a bug that some encountered where with a specific key-combination the screen would go dark
1.30 just slightly adjusted locking behaviour specifically between the PLL and AKM upsampling chip (no impact to sound or anything not involving the akm chip)

Just bug fixes, it won't impact USB, and you still won't get 30.12 working properly with AMD

[GoldenOne]

18 hours ago, ThenewGearPPK said:

I re-flashed to 30.14 firmware & now everything works.
 But why absolutely no music with 30.12?

As mentioned earlier, if you have an AMD source you need to stick with 30.14
Only switch to 30.12 if you have an intel source

[GoldenOne]

12 minutes ago, Roasty said:

I have a workaround for people running their core on AMD machines.

 

installed Windows 10 on a NUC 

loaded 30,12 firmware

HQPlayer NAA running on the NUC

settings - PCM1.536 on HQPlayer core AMD machine

 

can confirm, 1.536/1.4112M works through a Windows machine as NAA.

Nuc is an Intel device

[GoldenOne]

In case anyone was curious how the 1-bit converter in the may does (the may has a separate DAC entirely for DSD), the answer is pretty damn good....


Harmonics are crazy low. This is just 44.1khz into the may with 'OS DSD' selected

[GoldenOne]

Just now, luisma said:

Do you know if there is any firmware updates in the near future? where can you download these?

I don't think so.
If you need to update USB firmware the download page on the kitsune site will have the latest one.

The DAC itself there isn't any updates.

[GoldenOne]

3 hours ago, Extreme_Boky said:

 

May takes a bit of time to lock. I think they know this is an issue and I also think this is why they released firmware version 1.30. 

I am not sure if this firmware can be upgraded from 1.20 to 1.30; the jitter reduction and in particular the  reduction in time required to lock onto the incoming data stream would be definitely a welcome improvement.

 

Re remote, I have the same issue. I checked the remote control PCB traces, IR LED connections, confirmed that the green ceramic resonator (cheap oscillator) works correctly, ensured that the bottom plate does not touch the PCB (because it sits sooo close to the PCB), checked the batteries and battery sockets. It was all good - yet, the remote would most of the time just refuse to work. Then I got a bit worried because I realised that the issue is with the May DAC itself. This was a big disappointment, I must say. 

 

It turns out that I can get the remote to work if I cycle the mains switch at the back of the power supply unit (cycle the stand-by power  supply rails; cycling the power on the DAC front panel does not help....) The remote will then work for a while and then stop again.... so another power cycle will be required.

 

 

 

 

The firmware does not change dac behaviour or pll locking time. 

 

The locking time is just part of the accuracy of the pll. Stronger pll takes longer to lock. Just how it is. 

 

If it's an issue then it's recommended to use usb or turn pll off. 

 

1.30 doesn't alter anything 

[GoldenOne]

2 minutes ago, ThenewGearPPK said:

 

Try to upsample with the May to 768k with your software music player.

 

This. 

 

Definitely worth comparing the dacs on even footing. 

Else you're not really comparing the dacs and more just comparing if you prefer nos vs os or what os filter you like. 

 

Upsampling using the same filter for both will allow you to hear just the dac differences. And then look at what you like best for reconstruction later. 

 

Wavelight and May NOS are also fairly different so even having both on NOS will not actually be identical. 

 

May NOS has some analog reconstruction whereas wavelight has no analog lpf, but also overshoots on samples a bit. 

 

 

[GoldenOne]

Personally I've never found usb or power cables to make a difference when testing properly other than in one situation where an unshielded interconnect was picking up some interference from the power cable. But answer there is unshielded rca isn't a great idea. 

 

Can't hurt to keep stuff the same though. 

 

But different reconstruction approaches do make a big difference. May (or any dac) in NOS vs Internal OS vs HQP OS all sound very different (with HQP os being my favourite by some margin). 

 

Plus worth giving hqp os to dsd a go with the may given as the 1-bit converter is some people's preference. 

 

Not much point on the rockna though cause it doesn't convert true dsd. 

[GoldenOne]

1 minute ago, BrownMagic said:

In all fairness, one can only compare two boxes. Here the comparison is with a Rockna and a May. Not many are going to get into the details on how one’s tech is to the other. 

Asking one to bring an over sampling Software into the mix is not feasible for many. So if Rockna sounds better

on a particular mode compared to the May then it still does mean that the Rockna sounds better than the May according to @Roasty

No need for upsampling software or anything. 

 

Roon for example it's just a toggle. 

 

Or, could just have both in nos or both in OS to get closer to a fair comparison. 

 

For most delta sigma dacs the differences in reconstruction filters will be so slight that it doesn't matter for comparison. 

But NOS vs OS is a massive difference to the point where it could arguably outweigh differences between dacs. So quite important to not have that in the way

[GoldenOne]

1 minute ago, Extreme_Boky said:

 

There's a huge difference in sound between different power cables. I use them to fine-tune the end result; custom tailor the sound to individual tastes and the rest of the system. You should just try multistrand "standard" cable vs solid core pure copper - I guarantee you'll be amazed at the difference in sound.

 

You should also consider doing the listening tests not only with the headphones, but with the speakers as well. This will flash out the differences (in everything) much easier.

I have several higher end power cords here and have done comparisons. But I just use them for aesthetics. 

[GoldenOne]

1 hour ago, drjimwillie said:

Hi @GoldenOne

Thank you very much for the time and effort to put together those two incredible reviews on the May. 
Will you please say a few more words about the difference between the L2 and the KTE?

when you say all else is the same but it is 2% snappier, are you saying that it is 2% snappier on percussive elements only or is it on everything?

Could you tell the difference if listening to a quartet?

could you tell the difference if listening to rock?

I listen to everything, but mostly rock.

Thank you 

DJW

I could only tell the difference when listening to percussive elements or electronic type music. 
With things like vocals, strings, or other more 'continuous' elements I could not tell a difference.

And to be clear, the difference is very minor. It's not at all something I'd be able to pick out without the two side by side and listening careful. You're not 'missing out' by going L2. But if the $800 jump isn't an issue for you then there is a slight improvement to be had

[GoldenOne]

42 minutes ago, YALA said:

It isn't a DDC, it is just an hdmi to i2s converter and it works very fine.

That is a ddc

[GoldenOne]

7 hours ago, YALA said:

I know it is a DDC but not in the sense of a reclocker such as the denafrips ddc gaia which does have an i2s output but which does not convert an hdmi signal. If you know of any other high end hdmi / i2s converter, let me know because I'm a buyer.

Any format converter such as that inherently has to reclock as well. It unfortunately isn't possible to convert without retransmission, which then means it needs it's own clock which unfortunately for a cheap device like that is most likely going to be awful. 

 

What device are you actually feeding the may from? It's likely going to be easier to address that. Feeding it from an hdmi out is a little unusual

[GoldenOne]

18 minutes ago, JohnDonaldson said:

At the suggestion of Tim at Kitsune I recently installed a Ableconn PUSB31P2A PCIE USB card in my AMD X570 based motherboard in the hope it would resolve the problems with Holo firmware 30.12. My May now happily plays at 1.536 mhz. While the cards may not be of Audiophile level, I think it is at least of the same quality of the typical motherboard USB port and it only costs $34 at Amazon.

Many thanks to Tim for great customer support.

I've ordered one of these to test in my AMD system as well.
For those buying though, make sure to get that specific model number as it has the ASM1142 chipset, the newer one with the ASM3142 may not work.
 

[GoldenOne]

8 hours ago, Gavin1977 said:

Any know issues with JCAT USB Xe?

Pretty sure the jcat card won't do 1.5mhz. Someone earlier in the thread tried it I think (might have been in hqp thread I'm not sure) 

[GoldenOne]

14 hours ago, ted_b said:

As we’ve discussed before, those AMD server users who use HQPlayer can get 32fs by getting an inexpensive Intel NAA too. 

This is probably the easiest way, and what I'm currently doing. 

 

Small fanless Intel mini pc from 'mele' was £150 on amazon. 

Got roon Bridge and hqp naa running seamlessly side by side. 

 

And if anyone is Concerned about usb quality they can slap an Igalvanic Inbetween too

[GoldenOne]

15 hours ago, Extreme_Boky said:

The cards use a switching-mode power supply converter - terrible for audio - their implementation on these particular cards is especially quite bad. 

 

 

Can always add an Igalvanic or something if concerned. 

[GoldenOne]

1 hour ago, darioG said:

I have a Shanti powering the PI2AES and when use I2S to May L2 takes ages to lock and then when it does, doesn't always stay locked? If the PI2AES has not got as much jitter as denafrips dds, why is the May not locking as fast as when May is connect to my ollo USB sig and Shanti? 

Are you definitely using the correct I2S pinout?

Sounds like there is an issue with the connection, or your PSU is not providing enough power to the pi2aes. What voltage and current rating is the PSU?

For Pi2AES the I2S input on may should be set to "ALT1" layout