Topic: Homebrew 24/96 FLAC recorder

[spott]

Holy crap      , just saw this thing on bt.etree.org... is Burris on here?
-------------------------------------------------

http://bt.etree.org/details.php?id=18130

I use a little homebrew thing I made. Its got a pc104 266mhz pentium mmx, pdaudiocf input, 2.5" hd, lcd, buttons, etc... Runs linux with python and the heavy lifting done by a c module. It does 24/96 with on the fly flac compression. By using flac as my file format, I get twice the disk space, unlimited file length, up to 8 tracks in one file, and it takes half as long to transfer into my computer later. It has a 100 mbit ethernet jack and I ftp the data off, takes about 15 minutes for a typical concert. I've been meaning to make a smaller less junky case, fix all the bugs, do 4 tracks (266 mmx not burly enough to flac 4 tracks in real time though), have an 802.11 wireless remote so I can start/stop and monitor levels from the front row, and uh basically finish the thing, but I haven't had time, motivation, or especially money to do any work on it in 2 years.

Here's a few blurry phonecam pics, on the last you can see the stack of PCMCIA cage top, mobo in the middle, and disk drive on the bottom.

[JackoRoses]

Holy crap      , just saw this thing on bt.etree.org... is Burris on here?
-------------------------------------------------

http://bt.etree.org/details.php?id=18130

I use a little homebrew thing I made. Its got a pc104 266mhz pentium mmx, pdaudiocf input, 2.5" hd, lcd, buttons, etc... Runs linux with python and the heavy lifting done by a c module. It does 24/96 with on the fly flac compression. By using flac as my file format, I get twice the disk space, unlimited file length, up to 8 tracks in one file, and it takes half as long to transfer into my computer later. It has a 100 mbit ethernet jack and I ftp the data off, takes about 15 minutes for a typical concert. I've been meaning to make a smaller less junky case, fix all the bugs, do 4 tracks (266 mmx not burly enough to flac 4 tracks in real time though), have an 802.11 wireless remote so I can start/stop and monitor levels from the front row, and uh basically finish the thing, but I haven't had time, motivation, or especially money to do any work on it in 2 years.

Here's a few blurry phonecam pics, on the last you can see the stack of PCMCIA cage top, mobo in the middle, and disk drive on the bottom.

wow that thing looks like it rocks

[greenone]

Man, and I'd have been squeamish just replacing the trim pots on the V2.

[Ed.]

damn, nice.

[bconnolly]

That's really really damn cool.  A *nix nerd has his place in any hobby it seems.

[johnw]

Looks like he is using the Core Sound PDAudio for his digi-in. I wonder if he hacked the drivers or if Len makes them available for Linux. Either way, it's pretty impressive.

[sleepypedro]

can anyone mirror the pix here?  the site's down as far as i can tell.

slashdotted after only 7 views from ts.com members, nice

[bconnolly]

can anyone mirror the pix here?  the site's down as far as i can tell.

slashdotted after only 7 views from ts.com members, nice

Here you go:

[rePat]

Looks like he is using the Core Sound PDAudio for his digi-in. I wonder if he hacked the drivers or if Len makes them available for Linux. Either way, it's pretty impressive.

The PDAudio Linux drivers are built into ALSA, just need to be included at compile time.

[Simp-Dawg]

phrog built something similar into the empty casing of a da-p1, it used a digigram i think and ran linux, don't know if he put an lcd on it or not...

[BobW]

can anyone mirror the pix here?  the site's down as far as i can tell.

slashdotted after only 7 views from ts.com members, nice

Here you go:

No duct tape ?  Blasphemy !   

[macdaddy]

for open situations, that would be the wy to go...

i am still holding out hope that there will be a tascam-esque dsd machine that records to harddrive sooner than later.

[lds490]

Looks like he is using the Core Sound PDAudio for his digi-in. I wonder if he hacked the drivers or if Len makes them available for Linux. Either way, it's pretty impressive.

A Linux driver has been available for free on the Core Sounds web site for a couple of years. 

The latest version of Live2496 records direct-to-FLAC at 24/96 on a pda. 

[burris]

The whole thing is held together with gaffers tape.

[macdaddy]

The whole thing is held together with gaffers tape.

+t

it is still a great effort...

thanks for sharing, and it is great to see you around here.

welcome.

[Ed.]

The whole thing is held together with gaffers tape.

that, in itself, is a sign of quality

[johnw]

The whole thing is held together with gaffers tape.

+T 

[grayp]

i used to know a guy back in VA (I think his name was John Bartow) that tried something like this.  I remember seeing him at the Trey show in Pompano (02) after moving to florida and he was telling me about it.  I think that's the last time I saw him.  I wonder if this is the same guy or another generation from that effort.

[phr]

I've also been thinking of doing something like that, maybe with a Gumstix board (www.gumstix.com).  That's a 400 mhz XScale board with Bluetooth onboard, 64mb of ram, and Linux.  They have an SPDIF input board for it and I'd use a Core Sound a/d converter or something similar.  The missing piece would be a USB hi-speed host board with a hub, but I think that wouldn't be so hard to make.  I'd plug a USB hard drive into the USB board (and maybe a USB a/d converter if I didn't use the SPDIF) and control it from a PDA using Bluetooth.

I look at the commercial offerings (I have a PMD660) and they are all designed by idiots.  I think amateurs can do a much better job.

My PMD660 review is here:  http://www.nightsong.com/phr/pmd660.html

[it-goes-to-eleven]

I look at the commercial offerings (I have a PMD660) and they are all designed by idiots.  I think amateurs can do a much better job.

You haven't used a 722 yet?

For me, much comes down to time and asking the question: Do I want to make recordings or make recording gear?

[phr]

I haven't used a 722, which costs something like $2500.  It might be a great machine but when I speak of idiocy I'm referring to the kind of stuff described in my 660 review, just bonehead decisions by the designers or marketers that could have been done sensibly without making the recorder cost noticably more.  In the Microtrack's case, using AA cells instead of the captive li ion probably would have made it cost -less-.  With the 722, if they're saying avoiding obvious errors necessarily leads to a $2500 piece of gear, well, that's a different type of idiocy.  The 722 aims at a set of requirements that 660-like recorders make no attempt at all to fill, as opposed to attempting and failing for dumb reasons.  Marantz has been making portable cassette recorders for decades and they haven't figured out how to make a reliable set of battery contacts?   Also, the 722 appears to use way too much battery power, as if it keeps its HD going all the time rather than just spinning it up once in a while when an internal flash buffer gets full.  I don't think I could use a 722 the way I use my 660 without a heck of a lot more batteries.

The closest thing to sensible in solid state recorders seems to be the PDAudio which is a bit of a lash-up (not idiotic at all though, and I'll count it as a semi-amateur product since the company that made it is too small to have wing-tipped buffoons running around messing perfectly good things up--come to think of it, the same is probably also true for the 722).  I decided that if I go to 24 bit recording, then because of my long record-time requirements, I need something with a hard drive, otherwise I'd be considering the PDAudio pretty carefully.  Of course I could put the PDAudio CF card into a laptop, but then the amount of batteries needed starts getting cumbersome. 

Maybe a few people (I could be one of them) could get together and make some recorders using an open, published design and open software, and offer them for sale to tapers.  If it were me I'd probably start with the Gumstix board mentioned above, but doing a board completely from scratch (maybe with a floating point DSP so it could use the reference Vorbis encoder) might be within the realm of reason.

You are right though, I wasn't thinking of the 722 when I made that remark, it's just way more expensive than the stuff I've been looking at carefully.  The 660 was pretty close to my spending limit.

[Ed.]

the 722 uses camcorder batteries if i remember correctly.  and to me, a 6 hour camcorder battery would probably last the full 6 hours recording audio, when a mini dv cam will last that long and its recording audio and video onto tape.

not to mention they're easy to find and pretty cheap too.

[Evil Taper]

What about a nice laptop and an outboard pre/adc?  What you're saying is all the portables suck ass for one reason or another but a computer could be configured any way you like for a reasonable price.

[phr]

The 722 according to its FAQ uses about 1 amp of power (7.2 volts) when running full-out to a hard drive.  That means that with the largest camcorder pack (NP-F960) it can record about five hours.  In energy terms that pack is equivalent to about 12 high capacity NiMH AA cells.  This is tolerable but not great.  Use of the L-mount camcorder batteries was a very good choice though.  What would really be neat would be using two of them, like the Thinkpad 235 notebook computer did.  First of all that would let you record for ten hours with no swaps.  Second, you could hot-swap, i.e. change the batteries one at a time while the recorder is still running, so you could record nonstop for an unlimited amount of time, given enough batteries.

Also, the 722 doesn't support FLAC format, which just seems silly, since one of its recommended modes is recording to CF instead of HD.  It's like these companies are afraid of FLAC because it's a free program.  Maybe they can fix that in a firmware upgrade.

As for using a laptop, yes, it gives lots of recording flexibility.  The price for that is in size, fragility, complexity, power consumption, and speed.  To use a 660, plug in the mic, turn the 660 on, and press the record button.  To use a laptop, plug the SPDIF card into the laptop unless you're ok carrying the laptop with the card blob sticking out, plug the mic into the external pre/a-d, connect the a/d to the SPDIF card with a cable (fragile connector defeats the purpose of using XLR's on the mics), turn on the a/d, turn on the laptop, wait forever for the laptop to finish booting, log in to the laptop, launch some application and wait forever for it to come up, and finally you can record, maybe for 4 hours tops on a $100+ laptop battery pack (vs $30 for an off-brand NP-F960 pack on the 722, or $10 worth of NiMH cells in the 660 to record 8 hours).  The external a/d in some cases needs batteries of its own that you have to keep changing.  Plus of course you've got a cumbersome rig with all these components cabled together, that's sort of an attention magnet because it's an open laptop (I'm not trying to do "stealth" but don't want to create needless visual clutter).  I do think of using a laptop and might try it if I get a good a/d (Core Sound etc.) but the drawbacks (see above) are nothing to sneeze at.  A dedicated unit is a lot nicer.  If I win the lottery I'll certainly look into buying a 722.

I like to think that as flash memory gets cheaper (the Ipod Nano is the result of Samsung supplying Apple with flash chips at about 40% less per MB than the rest of us can get it for), hard disk recorders will become obsolete.  A 24/96 FLAC recording should use about 1.2 GB/hour and a 24/48 recording about half that.  An 8GB CF card will probably be below $200 within a year or so (based on the Ipod Nano breakdown) and will hold 6+ hours even at 24/96. 

Heck, it would be nice if 722-class machines had a USB 2.0 high speed host port to allow dumping files directly to a DVD burner without an intervening computer.  A notebook DVD burner is very compact and should easily fit in a carrying case with a 722.

[poorlyconditioned]

The 722 according to its FAQ uses about 1 amp of power (7.2 volts) when running full-out to a hard drive.  That means that with the largest camcorder pack (NP-F960) it can record about five hours.  In energy terms that pack is equivalent to about 12 high capacity NiMH AA cells.  This is tolerable but not great.  Use of the L-mount camcorder batteries was a very good choice though.  What would really be neat would be using two of them, like the Thinkpad 235 notebook computer did.  First of all that would let you record for ten hours with no swaps.  Second, you could hot-swap, i.e. change the batteries one at a time while the recorder is still running, so you could record nonstop for an unlimited amount of time, given enough batteries.

Also, the 722 doesn't support FLAC format, which just seems silly, since one of its recommended modes is recording to CF instead of HD.  It's like these companies are afraid of FLAC because it's a free program.  Maybe they can fix that in a firmware upgrade.

As for using a laptop, yes, it gives lots of recording flexibility.  The price for that is in size, fragility, complexity, power consumption, and speed.  To use a 660, plug in the mic, turn the 660 on, and press the record button.  To use a laptop, plug the SPDIF card into the laptop unless you're ok carrying the laptop with the card blob sticking out, plug the mic into the external pre/a-d, connect the a/d to the SPDIF card with a cable (fragile connector defeats the purpose of using XLR's on the mics), turn on the a/d, turn on the laptop, wait forever for the laptop to finish booting, log in to the laptop, launch some application and wait forever for it to come up, and finally you can record, maybe for 4 hours tops on a $100+ laptop battery pack (vs $30 for an off-brand NP-F960 pack on the 722, or $10 worth of NiMH cells in the 660 to record 8 hours).  The external a/d in some cases needs batteries of its own that you have to keep changing.  Plus of course you've got a cumbersome rig with all these components cabled together, that's sort of an attention magnet because it's an open laptop (I'm not trying to do "stealth" but don't want to create needless visual clutter).  I do think of using a laptop and might try it if I get a good a/d (Core Sound etc.) but the drawbacks (see above) are nothing to sneeze at.  A dedicated unit is a lot nicer.  If I win the lottery I'll certainly look into buying a 722.

I like to think that as flash memory gets cheaper (the Ipod Nano is the result of Samsung supplying Apple with flash chips at about 40% less per MB than the rest of us can get it for), hard disk recorders will become obsolete.  A 24/96 FLAC recording should use about 1.2 GB/hour and a 24/48 recording about half that.  An 8GB CF card will probably be below $200 within a year or so (based on the Ipod Nano breakdown) and will hold 6+ hours even at 24/96. 

Heck, it would be nice if 722-class machines had a USB 2.0 high speed host port to allow dumping files directly to a DVD burner without an intervening computer.  A notebook DVD burner is very compact and should easily fit in a carrying case with a 722.

I think you pretty much summed up my laptop experiences!  I've got a Presonus Firepod (eight mic/line inputs) with a firewire interface.  While I *love* the flexibility of running soundboard + one or two sets or room mics, this is much more complicated than a standalone recorder.  I only do this if I have AC power and if the venue is friendly, ie., not full of drunks spilling beer everywhere.  My usual venue is really laid back, has a nice sofa and table with AC power nearby, so I just sit down and wire up.

One of the biggest problems I've been having recently is with Windows ASIO drivers.  I plug in the firewire interface, but have trouble getting it to override the default sound system in my laptop.  I haven't completely figured out what is going on with this, and it often requires one or two reboots to get it right.  So, it is much harder and much less reliable than running a NJB3 or a dedicated HD recorder.

By the way, you mentioned Apple Nano, etc.  What I'm waiting for is a flash based recorder with line in (something like the Creative Nano), but with a removeable flash card.  Does such a thing exist?  Seems like it would be cheap to make, but I haven't seen one yet.

  Richard

[it-goes-to-eleven]

The 722 according to its FAQ uses about 1 amp of power (7.2 volts) when running full-out to a hard drive.

People seem to get more than that with phantom off. I recall reports of 6.  The manual specifies hard drive activity as 2 watts.  So, CF would be an option.

Use of the L-mount camcorder batteries was a very good choice though.  What would really be neat would be using two of them, like the Thinkpad 235 notebook computer did.  First of all that would let you record for ten hours with no swaps.  Second, you could hot-swap,

I'm pretty happy with their decision to use this series of battery.

The power cable on the 722 has 4 pins.  Two provide power to the electronics and two power the battery charge circuit.  It would be fairly easy to rig a cable for doing battery hot-swaps.

You have summarized the annoyances of laptop recording well.  I had my medium size vaio pretty well scripted for recording (on=record) but I still didn't like using it for the reasons you mention.

Flac would be nice on the 722.. I am guessing that it is not a priority for the majority of their customers.  I have one 4 GB CF but haven't used it with the 722. An 8 would be nice. Will we see a smaller 722 that eliminates the drive in favor of flash?  Maybe half the size?

Heck, it would be nice if 722-class machines had a USB 2.0 high speed host port to allow dumping files directly to a DVD burner without an intervening computer.  A notebook DVD burner is very compact and should easily fit in a carrying case with a 722.

That would be nice..  Though the host interface would add a lot of complexity to the recorder.  One alternative is to record to disk and cf simultaneously and remove the cf to a laptop.

If I win the lottery I'll certainly look into buying a 722.

For me, it came down to wanting a V3 level pre/AD and needing a backend. Having done laptops, I didn't want to go that route.  Microtrack was always the cheap holy grail. But it turned out a bit monty pythonesque..

I love the 722.. (who says I can't commit!)

[phr]

Re flash 722: I doubt the HD is much of the space or power budget of a 722.  It's likely all preamp and computer stuff inside.  The HD is a 2.5" laptop HD and they could limit it to a 1.8" drive (same as in a hard disk ipod, currently available up to 60gb) to save a little more space.  But other gadgets like the old Archos jukeboxes were much smaller and used 2.5" drives.  And with some internal ram or flash buffering, they only need to spin up the HD once in a while to write out the buffer. 

Given the expense of flash memory, if you're going to record to flash at all and if you've got the cpu power, FLAC seems like a no-brainer to me.

Re removable-media consumer flash gizmos with line-in: there are a few.  The old Pogo Ripflash is the best known, recording mp3 to so-called smartmedia cards (obsolete digicam cards limited to 128mb capacity).  The Frontier Lab Nex II / Nex IA use CF cards and are nice on paper and are cheap.  I have no idea about the recording quality.  There's a new model coming (Nexblack) and I'm planning to get one as a playback unit (it supports Ogg Vorbis).  Note I don't think any of these have manual level controls.  You could always take something like an NJB3 and replace the internal hard drive with a flash drive.

I feel like we're hijacking Burris's thread about his very cool sounding homebrew recorder with this discussion.  Maybe we should start a new thread.

[it-goes-to-eleven]

Re flash 722: I doubt the HD is much of the space or power budget of a 722.

The manual says 2 watts for the drive. That is probably accurate based on the raw specs of these drives.  Your previous figures assumed a total consumption of 1 amp at 7.2 volts. That is 7.2 watts. So the drive is approximately 25% of the power being consumed.  I'm not sure how much power a 4gb flash card would draw.  At some point I should do a flash test.

And with some internal ram or flash buffering, they only need to spin up the HD once in a while to write out the buffer. 

One challenge with that approach is that it takes 4.5 watts to spin up a drive and around 1.6 watts to write to it once it is spinning.  So for a given amount of buffer and recording rate, there is a sweet spot on power optimization in terms of how often you spin the drive.  At 24/48, the buffers seem to flush every 6 seconds on the 722.

The other problem with more buffering and writing later is the impact on other activities. These are fairly low power CPUs, though I'm not sure what CPU the 722 uses. So coordinating a thread that writes out a significant amount of recorded data is complex and increases the risk of data loss.  SD says that they have just enough power reserve in the power supply to flush the in-memory content to disk if the power is lost. I would hate to lose that capability.