Crackles on UA-25 and US-224

[Ivor]

Hi All!
Trying ASIO4ALL as a sneaky way of using multiple soundcards in the ASIO environment of Cubase VST 5.1R1 on an Acer Travelmate 640 lappy (1GHz P3). Two high speed USB ports provided via a PS2 port powered PCMCIA card. OS = XP Pro.

Using new Edirol UA-25 USB interface and the Tascam US-224. The dedicated ASIO apis that come with the cards work perfectly (albeit one card at a time) and they both boast low latency WDMs for use with Cakewalk (so you imagine the WDMs are reasonably well written).

ASIO4ALL sees both the Edirol and the Tascam. Oddly, the US-224 output and its pin was showing as 'Beyond Logic' and ASIO4ALL was reporting an overload. I disabled the US-224 output and the problem went away (I wasn't too worried about that because I was going to use the UA-25 as my line out anyway).

Default settings gave random crackles on playback but appears OK on record. Bursts of midi activity such as using the jog/shuttle wheel on the US-224 would be followed half a second later by a bunch of crackles but just leaving the US-224 alone would not guarantee that crackles would not appear.

All buffer settings tried. Larger buffers meant less crackles or at least it seemed that it took longer for the first crackles to appear. Suggestions please.

Someone's going to ask about how I'm going to sync these two cards together if I get this working and I'm probably going to go 'Doh' because although they both have digital I/O the US-224 is SPDIF and the UA-25 is optical Oh well, can't be clever all the time

[Michael Tippach]

In the Windows control panel/Sounds and Multimedia, is there, by chance, the US-224 listed as the default playback device?

If so, this would be the device the system would use when the MS GS Software Wavetable Synth is started, which would explain why the outputs of the US-224 are being flagged with an error. You can disable the MSGSSWWTS in the Cubase MIDI options.

Do you get the crackles also if you _only_ enable the UA-25?

[Ivor]

Hi Michael,

Yip, spotted the MSGSSWWTS tip before posting and its not that.

After much fiddling last night:
I now have all devices and pins showing as enabled and running in the offline panel. Open Cubase then the ASIO panel from within Cubase and all devices and pins still show as enabled and running except the overload indicator is flashing at me.

What I did was this:
I selected each device individually in Cubase and opened the ASIO panel and ensured that the buffer size/latency figures for each device were the same as the setting used in ASIO4ALL. Then I selected the ASIO4ALL device in Cubase, opened up the ASIO panel and noticed that I no longer had 'beyond logic' devices, all were showing as running - still unfortunately with overload flashing.

I tried 16bit and 24 bit, no difference. 44.1 or 48, no difference. I selected 'global disable' of monitoring, no difference (these devices boast input monitoring so I disabled in case that was the problem). Always saving, shutting down and restarting Cubase between changes.

What I noticed was that there is an offline ASIO panel for the US-224 that declares certain things about the device that you cannot change in the ASIO panel itself. For example, it tells you it's running at 24 bit, 44.1KHz but with no way of changing it. It gets that information from somewhere and I reckon it gets it from the ASIO compatible application, e.g. Cubase, when it gets started and remembers it when it stops. Only in the Cubase audio setup window are there boxes to select bit depth and sample rate.

There is no offline panel for the UA-25 and the panel that is presented within Cubase only has a latency slider.

I wish to ask:
Is it possible that there is flash in the device hardware that gets written to by the native ASIO drivers that doesn't get written to with ASIO4ALL?

Does Cubase 'communicate' these settings across the ASIO layer to the WDM layer, or does ASIO4ALL get told the settings by Cubase and ASIO4ALL configures the WDM device accordingly?

Is ASIO4ALL written to accomodate multiple devices under any circumstances? (probably not, I would expect both devices to at least be on same bit depth and sample rate - or are you indeed God? )

I very much appreciate your work with ASIO4ALL. You are highly praised by Martin Walker of Sound On Sound and if ASIO4ALL can provide the first multiple device interface for ASIO applications like Cubase then you will surely go down in history! (I wanted the US-224 for its control surface, but it has no phantom power for mics. I wanted the UA-25 for its phantom power. I wanted both of them because they were USB powered - nobody makes everything in one box, the world NEEDS multiple ASIO devices)

I hope that my questions might suggest a reason for my problem and you can help in finding the solution.

Best Regards, Ivor

PS Never tried JUST the UA-25 enabled in ASIO4ALL, I was always trying ASIO4ALL as the multicard solution. I can confirm that the Edirol and Tascam drivers are rock solid on their own.

[Ivor]

Hi Michael,

I can confirm that ASIO4ALL is rock solid when only ONE of the USB devices is enabled in the ASIO4ALL panel. This proves that ASIO4ALL is capable of communicating with one card at a time, but shows instant overload if both cards are enabled at the same time.

Default playback and record in Windows is set to the laptop's onboard device which is disabled in ASIO4ALL, midi set to the UA-25's midi port. All windows sounds turned off.

BTW, my standard test is to record one track, then play it back while recording a second track. Then play back the two recorded tracks while recording a third. Once it becomes stable doing this I will then try the full monty of recording more than one track at once.

[Michael Tippach]

Hi Ivor,

Even with all possible improvements that may be made to ASIO4ALL, without the devices being sample rate locked there will always be artefacts.

You could try to relax the buffer size a little (ASIO4ALL always uses the largest size of all devices enabled).

Temporarily, just to establish the general notion that it _might_ work once the devices are hard synced, you are encouraged to try the test version that I recently put up on:

www.asio4all.com/a4a23pre.exe

The changes in the audio engine that I made here may actually give your devices a little more headroom before the overload flashes.

[Ivor]

Hi Michael,

Many thanks for the response.

Further update on situation is that the problem 'appears' to be with the US-224.

A test I did last night was to try bringing the onboard AC97 device into the equation...
UA-25 + AC97 =
UA-25 + US-224 =
AC97 + US-224 =
UA-25 + AC97 + US-224 =

As soon as you enable the US-224, the wheels fall off.

I'll concentrate on getting the two USB devices working and try out your latest beta.

[Ivor]

Tried version 2.3 and it does work.

With buffer size of over 1500 samples, both the US-223 and the UA-25 co-exist without overload - so 2.3 is definitely on the right track

What is interesting though is that with a buffer size of 1500, the US-224 (the one that I reckon is the trouble maker) has a short and useable latency of around 50mS. Meanwhile, the UA-25 alongside it on the same buffer size has a latency of 700mS

Does this huge difference between the two devices in latencies for a given buffer size tell you something deep down that explains the problem?

BTW, I tried to use generic windows USB drivers on both these devices but they wouldn't play ball. They clearly have some magic dust between them and the OS that I'm sure explains the strangeness.