Jump to content

Amateur OB speaker builder and his ARC based digital system


Recommended Posts

My DSPeaker X4 has been misbehaving on the microphone input for measurements and calibration. They've diagnosed the problem over email as a simple resistor that needs replacing - mine was the first batch they ever made and in later models they use a higher power resistor in that location. So I have to go and drop it off to be fixed. The practical upshot of this is I lose my precious digital room and speaker correction. Anyone who's been here and listened to my system would know how dependent on DSP my OB speakers are and it's pretty much not worth listening to them without it. Numerous times I've tried to recreate the X4's magic in software to no avail - it always sounded compressed and lifeless. The thought of going without my speakers was too much though so I spent a day working on the software approach yet again, but with FIR convolution. This time the result actually turned out pretty darn good. I'm just using a dsp ladspa plugin with pulseaudio at the moment; it took me a while to realise that I had to make sure the filter wasn't too long or it starts misbehaving with pops and clicks (making it feel like vinyl hah.) This isn't a comprehensive solution though because I'm limited to just the one sample rate which means I need to resample everything (in pulseaudio with sox resample) to that one rate - that's precisely why the X4 is magic as I've discussed in other threads because it does everything at native resolution. I could load a different filter whenever I change sample rates but that's laborious and not really practical. I had a play with hqplayer loading the filter there and it seemed happy even though it was a different sample rate. It's not clear what it's doing to be happy with a 192kHz convolution filter when it's playing 44kHz sample rates, but it comes out at 44. It doesn't offer multiple convolution filter inputs for different sample rates. The thing is hqplayer is rather clunky as a media player so I'd need both it and roon making this software solution cost me quite a bit. Since it's just a temporising measure I think I can cope with resampling everything to 192.

Edited by Ittaku
  • Like 1
Link to comment
Share on other sites



2500 hours on my valves and it's testing time again. All the power amp valves are still performing very well and I suspect have plenty of life left. I did my usual rotation of valves that I do that brings those in the hottest region to the coolest region - i.e. closer to the rear cooling fans. Interestingly, though, I dropped yet another regulator tube in the preamp power supply. That's the second one that's gone below spec in that position. I wonder if there's a reason that particular one has gone twice. it's not any warmer or anything, and while it's still functioning, it's below minimal operating spec where it used to be 50% above spec. Mick said the regulator tubes should last 10,000 hours but these have only lasted 1000 each. It's a shame I don't have a way of testing the rectifier tube for anything except shorts and it's still okay after 2500 hours despite allegedly being good for only 2000. Below is a pic of the 7581A on the right failing to meet its spec.

 

IMG_20191024_105922.thumb.jpg.e4e6da47e05c0c229e81ded74823df00.jpg

Edited by Ittaku
  • Like 1
Link to comment
Share on other sites

9 minutes ago, Ittaku said:

2500 hours on my valves and it's testing time again.

Hmmm    that reminds me, I haven't tested my 6B4G SETs.  Haven't even looked up the rated lifetime for them.  Might have to assume they are similar to a 2A3.

 

16 minutes ago, Ittaku said:

Interestingly, though, I dropped yet another regulator tube in the preamp power supply. That's the second one that's gone below spec in that position. I wonder if there's a reason that particular one has gone twice. it's not any warmer or anything, and while it's still functioning, it's below minimal operating spec where it used to be 50% above spec. Mick said the regulator tubes should last 10,000 hours but these have only lasted 1000 each. It's a shame I don't have a way of testing the regulator tube for anything except shorts and it's still okay after 2500 hours despite allegedly being good for only 2000. Below is a pic of the 7581A on the right failing to meet its spec.

I'm a little confused.  So, a 7581A regulator tube has failed twice, but you also say "It's a shame I don't have a way of testing the regulator tube for anything except shorts and it's still okay after 2500 hours despite allegedly being good for only 2000"  Are there two regulator tubes in your amp perhaps?

 

Anyway, I would be trying to measure the operating point of the one that is failing (voltages, current, power dissipation).  It may be running off spec, or the design conditions might run it too hard (unlikely I suppose).

Link to comment
Share on other sites

1 minute ago, aussievintage said:

I'm a little confused.  So, a 7581A regulator tube has failed twice, but you also say "It's a shame I don't have a way of testing the regulator tube for anything except shorts and it's still okay after 2500 hours despite allegedly being good for only 2000"  Are there two regulator tubes in your amp perhaps?

I meant to say I have no way of testing the rectifier tubes :P Corrected in my post now.

Edited by Ittaku
Link to comment
Share on other sites

1 minute ago, aussievintage said:

Anyway, I would be trying to measure the operating point of the one that is failing (voltages, current, power dissipation).  It may be running off spec, or the design conditions might run it too hard (unlikely I suppose).

Would probably need to check directly at the preamp as the tubes started out measuring identically between left and right. I'm not sure I trust myself with those voltages.

Link to comment
Share on other sites



1 minute ago, Ittaku said:

Would probably need to check directly at the preamp as the tubes started out measuring identically between left and right. I'm not sure I trust myself with those voltages.

One hand behind your back!      That's a useful way to do it.  Clip one multimeter lead to the system common/ground, then measure with one hand holding one probe, the other behind your back.   If the circuit is 3 dimensional and there's a danger of shorting to other bits with the probe, use an insulated probe, or a second clip lead - which you apply with the power turned off, then stand back and turn it on, watching the meter.

Link to comment
Share on other sites

16 hours ago, Ittaku said:

For the very few people that know I moonlight in linux kernel hacking, here's an update:

https://ck-hack.blogspot.com/2019/10/linux-53-ck1-muqss-version-0195-for.html

You'd be surprised what people get up to in their other lives I think.     

 

Do you remember CP/M, back before IBM PCs and DOS?  I once acquired one of the first double sided mini-floppy disc drives and the computer manufacturer had not released an update to understand them.  So, I hacked the CP/M "driver" and rewrote the format program to get mine up and working.  Turns out, when the official release came through, my version actually stored more data, but was probably less efficient due to sector size choices I made.

  • Like 1
Link to comment
Share on other sites

7 hours ago, aussievintage said:

Do you remember CP/M

I used ZCPR3 instead of CP/M.

 

Those days there were quite a number of options for computing, I guess Linux came from those origins with people making things "work better".

I still have a couple of "Osborne"s and a KayPro luggable, provbably have disks somewhere too!

 

Do not do much of that these days nice to see someone doing so though.

Link to comment
Share on other sites



Got my X4 back and have been reworking my room correction again. It was still under warranty so that was nice. Now that I've had my speakers stable for an extended period, it's become apparent to me that my DSP is doing too much "speaker correction" in addition to the room correction. Whilst it's not necessarily a bad thing since the DSP is a permanent part of my system, I know that I'll get a better sound with less boosting and relying mostly on gentle taming of peaks instead. Now that I've got a good handle of my speakers' in-room response, I've had a long think about the balance of the drivers and think I can improve things a lot to require less speaker correction. So I ordered another $250 worth of resistors to tweak the tweeter and midrange crossovers. It won't be an extensive rework, only affecting balance. Because of the effect the resistors will have on impedance of the midranges below their crossover point, it will cause much more attenuation at their resonant frequency. This should lead to more power handling as the midranges are the rate-limiting part of the speakers - even though it gets deafeningly loud (by my standards) now already, and I do have loads more power to play with. Less excursion of the midranges at higher volumes should improve their linearity as well. Parts should come within 2 weeks for this next fine-tuning.

  • Like 1
Link to comment
Share on other sites

2 hours ago, Ittaku said:

Got mine in the mail. Here're the results.

 

Without filter:

MVIMG_20191030_102746.thumb.jpg.56a7a156d5bc36adf787b0c38d9a635d.jpg

 

With filter:

IMG_20191030_102813.thumb.jpg.32664782991886cc6759f98eb4638771.jpg

 

So at the very least, I can conclusively say these definitely filter out power line noise. Note this is not an endorsement that it makes any audible difference though. Assuming equipment has its power induced noise floor noise at -120dB, that means that in this example the baseline noise is at -145dB (25dB lower), and this change would have decreased it by 5dB to -150dB (30dB lower), while the power line noise floor would still be at -120dB.

From the Thor filter purchase thread. Interestingly it's a better filter than the Sequenced Power Interface I was using, but it doesn't have the other features that has. Maybe I can chain them for combined effect? Yes it does work that way.

Edited by Ittaku
Link to comment
Share on other sites

Got my replacement 7581A tube for the preamp regulator today. I also got myself a spare Sovtek 5AR4 rectifier since the last one worked so reliably and was theoretically meant to die a long time ago, so that I have a spare for if and when it does finally go.

Link to comment
Share on other sites

Now, I can't decide if I need to buy myself a small signal tube tester as well. It would be nice to have but costs even more than the power tube tester and it costs nowhere near as much for regular off-the-shelf small signal tubes, but they underperform in subtle ways without failing so it's hard to know. ARC say the 6H30Pis have 4000 hours of life in them so I think with careful management I can make them last ages too, but I'm not sure it's worth it to get a tester that costs almost $1500. Thing is, I can use it to test the expensive globe tubes in my preamp as well, but they should be good for 10,000 hours. Hmm, maybe I'll consider getting one when I'm around 4000 hours on the small signal tubes in the power amps.

Link to comment
Share on other sites

25 minutes ago, Ittaku said:

Now, I can't decide if I need to buy myself a small signal tube tester as well. It would be nice to have but costs even more than the power tube tester and it costs nowhere near as much for regular off-the-shelf small signal tubes, but they underperform in subtle ways without failing so it's hard to know. ARC say the 6H30Pis have 4000 hours of life in them so I think with careful management I can make them last ages too, but I'm not sure it's worth it to get a tester that costs almost $1500. Thing is, I can use it to test the expensive globe tubes in my preamp as well, but they should be good for 10,000 hours. Hmm, maybe I'll consider getting one when I'm around 4000 hours on the small signal tubes in the power amps.

You could just trust your ears.  Do an AB test against some new spares, and if gain and sound doesn't change significantly, keep putting the old ones back until next time.

Link to comment
Share on other sites



Just now, aussievintage said:

You could just trust your ears.  Do an AB test against some new spares, and if gain and sound doesn't change significantly, keep putting the old ones back until next time.

That sounds like a good idea, I guess I can trust my left ear ?

Link to comment
Share on other sites

You may have seen an OpenDRC-DA8 in some of my latest pictures. The reason for getting this device was to deal with a latency issue my DAC has. In my system, digital comes into the X4 and then is split into two outputs - the main speaker output for >= 30Hz and the subwoofer output for <= 30Hz. Ever since I got the MSB DAC, I've been using the digital output from the main speaker output component to then do the analogue conversion, and the subwoofer output directly. There's been an issue with this approach as the DAC adds a variable amount of latency due to its extreme oversampling, making the mains delayed from the subwoofer. Now this is normally a good thing since crossovers often add delay to the mains to time and phase align with the laggy subwoofer output. Except that the DAC adds quite a lot of latency, and it varies according to the incoming sample rate. I purchased the OpenDRC to experiment with adding enough delay to the subwoofer output to match it. The reason for OpenDRC instead of the all in one mindsp is they don't offer enough delay! The OpenDRC can have up to 3 seconds' delay, and MSB told me there can be up to 500ms delay worst case scenario (though that is only with ultra low sample rates below even CD resolution) whilst the mindsp can only add about 15ms.

 

I ran into an issue even before I'd started though with the openDRC because it takes coaxial digital in, and the subwoofer output from the X4 only does analogue out. So I bought myself a $20 ADC (no kidding) to convert the subwoofer output back to digital. Now you're probably thinking I'm nuts to use such cheap garbage with such an expensive system, but the reality is that even a cheap ADC is more than adequate for subwoofer duty since it only has to do music up to 30Hz. I even played with the ADC and took measurements to check its linearity and couldn't actually fault it! That's right, even cheap ADCs have excellent frequency responses and linearity these days.

 

So now the signal goes

PC -> Singxer SU-6 -> DSPeaker X4 ->

->MSB DAC -> preamp -> power amp

->ADC -> OpenDRC -> preamp -> subwoofer

 

You may recall I have a preamp that takes 4 channels in and out for this to work. For the MSB DAC to work prior to this, I have to use the one shared output and just "soft crossover" the subwoofer by cutting the output below 30Hz dramatically and then boosting the subwoofer output dramatically to compensate. It was never a satisfactory solution and bugged me that I wasn't really using the X4's crossover component. With enough experimenting I discovered the MSB DAC added 42.5ms delay when the incoming sample rate was 88/96kHz, and 170ms when it was 44/48kHz. So I created two profiles for the OpenDRC but since virtually all my music is 88/96 and I can upsample everything else to that in the PC, I only really need the one profile in use.

 

So how does it sound? Interestingly the frequency response graphs look almost identical, but the improvement in ultra low bass authority is astounding. I had a feeling I wasn't getting the most out of my bottom end with the previous approach but this was really illustrative. I can't really tell if it's because the woofers are no longer doing the frequencies below 30Hz or if the subwoofer was being choked, or a combination of both. Either way, it's a fine result that I'm happy with for now.

 

Here's a picture of this crazy setup with the ridiculous cheap ADC that is too small to even sit flat.

 

VLROeu0.jpg

Edited by Ittaku
Link to comment
Share on other sites

It also turns out the miniDSP 2x4HD can be configured for a maximum delay of 80ms. That's enough for 88/96 sample rates but not enough for 44/48 in my configuration so the OpenDRC is still the better solution for me for now. The regular 2x4 can only do 27ms max.

Link to comment
Share on other sites

Interesting experience with the Thor filtered power board. it has two filtered banks, and the sockets on each bank share the same filtering. Above is a picture of the line noise without connecting my humidifier, and below with it connected to the same bank. You can see how the humidifier injects noise into the same circuit - even when it's off. Plugging it into the other bank the extra noise doesn't show up. Note that the baseline noise has to be quite low to pick up this amount of extra noise. On weekdays it's much higher so it wouldn't show up, so I can only do these experiments now. The interesting part is that the linear power supply I use to run the ADC for the subwoofer injects just as much noise as this humidifier, and yet the SMPS that came with the OpenDRC injects nothing back at all, so it's nice to see it comes with a good PSU as an audio component. I don't have the equipment to compare noise coming out of the power supplies though. Of course this tells me nothing of whether any of this translates into better audio performance but it's nice to minimise potential noise.

 

MVIMG_20191103_154728.thumb.jpg.83893a648f5b72a31282ad8de357db36.jpgMVIMG_20191103_154735.thumb.jpg.a4aff144cbabd407fff707ac6839127c.jpg

  • Like 1
Link to comment
Share on other sites

3 minutes ago, Ittaku said:

The interesting part is that the linear power supply I use to run the ADC for the subwoofer injects just as much noise as this humidifier, and yet the SMPS that came with the OpenDRC injects nothing back at all, so it's nice to see it comes with a good PSU as an audio component.

Yes.  I recall a recent discussion on SNA about linear vs SMPS, and how just replacing an SMPS with a linear was no guarantee of an improvement. It depends on the quality and design.

Link to comment
Share on other sites



Just now, aussievintage said:

Yes.  I recall a recent discussion on SNA about linear vs SMPS, and how just replacing an SMPS with a linear was no guarantee of an improvement. It depends on the quality and design.

Indeed, but I thought it was more on the downstream side, but clearly it's both sides!

Link to comment
Share on other sites

Did some more experiments and the linear power supply doesn't introduce any noise back into the power when nothing is drawing power from it, but with the OpenDRC connected to it, it injects noise back in. I have two different inexpensive LSPUs and they both performed the same. The SMPS doesn't inject noise when it's powering the OpenDRC or not. Again this doesn't tell me what's going on downstream, but given it doesn't isolate the mains from noise from the devices, I'd be surprised if it performed well in that direction now and don't feel inclined to even use it any more.

Edited by Ittaku
Link to comment
Share on other sites

To be honest, I am not sure the issue of noise on the mains is as big as some feel it is.  Power supplies, by design, always have huge filtering sections on the final DC output, and you can always add mains filters on the upstream side.   SMPS or linear, I use whatever I have that does the job.

Link to comment
Share on other sites

8 minutes ago, aussievintage said:

To be honest, I am not sure the issue of noise on the mains is as big as some feel it is.  Power supplies, by design, always have huge filtering sections on the final DC output, and you can always add mains filters on the upstream side.   SMPS or linear, I use whatever I have that does the job.

I'm quite aware and in agreement. In fact I've never once heard any kind of power conditioning, treatment, filtering, cable, or otherwise make an audible difference on my system whatsoever. That said, I'd have to classify the power supplies in my equipment as being in the insane price category so I'd hope so. Nonetheless, it's good to know I'm not doing any potential harm.

Edited by Ittaku
Link to comment
Share on other sites

Sold off the SU-6 as someone just happened to be WTB one. Since I couldn't find it offered any audible advantage I didn't see the point in holding onto it. It did improve the sound going to the Topping DAC, but it was subtle, and it limited sample rate to 384kHz max by doing so, and the improvement of oversampling to 768kHz was actually worth more to me. I don't know if the improvement in sound was due to USB isolation, lower jitter, or both.

Link to comment
Share on other sites



  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...
To Top