As founder and ex-chief design engineer of my company, and firmly entrenched in, and very happy with, 20th Century analog technology, it took a veritable Act of God to even make me acknowledge audio in digital form. But when I found that even I, with no code-writing skills at all, could design DSP-based products, I became a semi-convert overnight. "Semi," mind you, because I still like a simplistic approach to any problem, be it electronics, plumbing or women. But what Analog Devices has brought to the table with their Sigma and SHARC development systems is nothing short of magic. It takes a recent-graduate EE to make the conversion from a Sigma-based design to hardware, and to provide the conversion from Sigma's files to actual code that runs the project, but it is just so damned easy... and cheap!
I have "breadboarded" any number of little personal projects using the Sigma development module, and as long as one is content to tying-up an old laptop to run the circuit, it's the proverbial Duck Soup. I've got two of those development platforms, an original, which works with the high-horsepower chips, and a newer one that goes with the more recent low-cost chips with all the ADC stuff built in.
Despite the fact that I only just came up with an all-analog design using more than thirty(!) 2N3904/06 transistors (at 2¢ each), I do acknowledge the digital approach as the wave of the future. We all should be playing with this.
KNOP lathe DIY build
Moderators: piaptk, tragwag, Steve E., Aussie0zborn
-
EpicenterBryan
- Posts: 738
- Joined: Sun Aug 10, 2014 9:01 pm
- Location: Eugene, OR USA
Re: KNOP lathe DIY build
I downloaded the software and watched a few videos. Amazing!grooveguy wrote:We all should be playing with this.
I'm thinking this might be the ticket for an automatic pitch controller. It could even do the look ahead / delay function, in addition to HF limiting, iRIAA and probably depth control too. Crazy.
What do you think of this board set I found? Enough options on this set?
http://store3.sure-electronics.com/adau1701-audio-digital-signal-processor-kernel-board
They also sell them on E-bay... about $65 for the core board, programmer, cables and audio in / out board. And it runs stand alone once programmed.
-
grooveguy
- Posts: 467
- Joined: Thu Jun 22, 2006 5:49 pm
- Location: Brea, California (a few miles from Disneyland)
- Contact:
Re: KNOP lathe DIY build
Hey, Bryan,
Don't know what to tell you on that one. It's not the board I have, which is pictured below and comes directly from Analog Devices. The one you linked me to appears to be an aftermarket affair of some sort, and may be every bit as good as 'the real thing,' but I can't vouch for it.
The A/D (Analog Devices) approach is a two-board one. The one that goes to your USB port is the small guy, which connects with the ADAU1701 board with a JTAG connector over the 10-pin ribbon cable. This same small board can plug into the mother board on the products we have Sigma chips on, so you can run from the Sigma program directly instead of having to compile code. This comes in handy when you want to change parameters temporarily just to see how it sounds. Once the Sigma schematic is updated, then another guy turns it into code and blasts it into memory to load at boot.
The other A/D board has a pot, a Class D output amp and some other bells and whistles I've never used. The ADAU1701 is a remarkable chip; we have several professional products using it as the main go-to processing part. This product:
http://inovonicsbroadcast.com/product/223
uses just one ADAU1701 and a PIC to do everything. I kept expecting to run out of headroom, but even with all the functions going on here, it's still under 70% of capacity.
The Sigma interface can be a bit cantankerous. I started with Sigma under Windows XP, with most of my designs done under Vista and Win7. I asked our programmer for a recommendation for a new computer, as my old Asus 'mini' was very slow, what with all the MS updates. He picked out a Dell with Win10 that he thought was best for the purpose. Bottom line: it does not like Sigma, and the USB interface crashed whenever you went into the program to change a frequency, gain or whatever. So I continue to run Sigma on the old Asus, which I "upgraded" to Win10, and the Dell is now on my desk in the office that I'm using right now. The old Asus is slow but it works fine. So evidently there are some computers that just don't like the Sigma interface, the A/D one anyway.
There's an active and helpful A/D forum/group for Sigma, and they've bailed me out of more than one problem. No help with that Dell+USB issue, but when you're selling hundreds of thousands of dollars of a product, a $600 dollar computer just can't become an impediment, nor can it justify a shortcult to an alternative development platform, which may, or may not, mesh well with the forum/support group.
Don't know what to tell you on that one. It's not the board I have, which is pictured below and comes directly from Analog Devices. The one you linked me to appears to be an aftermarket affair of some sort, and may be every bit as good as 'the real thing,' but I can't vouch for it.
The A/D (Analog Devices) approach is a two-board one. The one that goes to your USB port is the small guy, which connects with the ADAU1701 board with a JTAG connector over the 10-pin ribbon cable. This same small board can plug into the mother board on the products we have Sigma chips on, so you can run from the Sigma program directly instead of having to compile code. This comes in handy when you want to change parameters temporarily just to see how it sounds. Once the Sigma schematic is updated, then another guy turns it into code and blasts it into memory to load at boot.
The other A/D board has a pot, a Class D output amp and some other bells and whistles I've never used. The ADAU1701 is a remarkable chip; we have several professional products using it as the main go-to processing part. This product:
http://inovonicsbroadcast.com/product/223
uses just one ADAU1701 and a PIC to do everything. I kept expecting to run out of headroom, but even with all the functions going on here, it's still under 70% of capacity.
The Sigma interface can be a bit cantankerous. I started with Sigma under Windows XP, with most of my designs done under Vista and Win7. I asked our programmer for a recommendation for a new computer, as my old Asus 'mini' was very slow, what with all the MS updates. He picked out a Dell with Win10 that he thought was best for the purpose. Bottom line: it does not like Sigma, and the USB interface crashed whenever you went into the program to change a frequency, gain or whatever. So I continue to run Sigma on the old Asus, which I "upgraded" to Win10, and the Dell is now on my desk in the office that I'm using right now. The old Asus is slow but it works fine. So evidently there are some computers that just don't like the Sigma interface, the A/D one anyway.
There's an active and helpful A/D forum/group for Sigma, and they've bailed me out of more than one problem. No help with that Dell+USB issue, but when you're selling hundreds of thousands of dollars of a product, a $600 dollar computer just can't become an impediment, nor can it justify a shortcult to an alternative development platform, which may, or may not, mesh well with the forum/support group.
You do not have the required permissions to view the files attached to this post.
-
EpicenterBryan
- Posts: 738
- Joined: Sun Aug 10, 2014 9:01 pm
- Location: Eugene, OR USA
Re: KNOP lathe DIY build
I see. What you show is what they call "EVAL-ADAU1701MINIZ - Limited Capability Evaluation Board : Cost $195". The small board (JTAG) I think people also call a USBi - your interface between the studio software for debug / tweaks and is also your programmer for the EEPROM. The actual USBi board is included in your kit and on it's own is listed as $49 but requires purchase of another evaluation board...grooveguy wrote:It's not the board I have, which is pictured below and comes directly from Analog Devices.
Anyway, from the link I listed before they have a USBi board compatible with SigmaStudio that works with the core. It's one of the things I noted that came to a total of $65.
Here is what I have on order (like I need another project):
ADAU1701 Audio Digital Signal Processor Kernel Board: $19.90
Interface Extension Kit (cable and board with connectors for audio in / out): $9.99
Open Sourced Programmer For Sigma Studio (USBi): $14.90
In-Circuit Programmer for ADAU1701 (alternate way to program from Hex file I guess): $19.99
I'm not sure if I need the in circuit programmer since the "Open Sourced Programmer For Sigma Studio" should also be able to program but I ordered both anyway...
Boom. I'll let you know how it works out.... I checked out your Multimode box video. Very cool. So The Pic runs the display / menu / data entry and passes that stuff to the ADAU1701?
Bryan
-
grooveguy
- Posts: 467
- Joined: Thu Jun 22, 2006 5:49 pm
- Location: Brea, California (a few miles from Disneyland)
- Contact:
Re: KNOP lathe DIY build
Hi, Bryan,
Yes, that's pretty much how it works. The Sigma code is in RAM, and the PIC blasts it into the chip on boot. I find the whole concept mind-boggling. So easy and so inexpensive. Of course there's the investment in the development stuff and the cost of a 'hip' engineer who can understand all this, but... just saying.
Yes, that's pretty much how it works. The Sigma code is in RAM, and the PIC blasts it into the chip on boot. I find the whole concept mind-boggling. So easy and so inexpensive. Of course there's the investment in the development stuff and the cost of a 'hip' engineer who can understand all this, but... just saying.
Re: KNOP lathe DIY build
An affordable hardware DSP solution looks very interesting. I wonder if the latency can get low enough to use in a studio setup as a realtime effects unit or VA synth?
Did some 3D work today,
Got the alu torque tube down to 0,7 gram from 1,05 gram, I hope I didn't shave too much weight. Also got an alu printed 'cone' thing going. That will weight about 0,3 gram per driver, plus glue weight. I also ordered some of the perfume funnel things. I will try and see what works best.
I'm pretty excited to try out some of this new stuff, too bad I don't have all this fancy 3D printer equipment at home.
Also noticed that some AC noise is getting through to my amp. Will need a power conditioner I think.
Did some 3D work today,
Got the alu torque tube down to 0,7 gram from 1,05 gram, I hope I didn't shave too much weight. Also got an alu printed 'cone' thing going. That will weight about 0,3 gram per driver, plus glue weight. I also ordered some of the perfume funnel things. I will try and see what works best.
I'm pretty excited to try out some of this new stuff, too bad I don't have all this fancy 3D printer equipment at home.
Also noticed that some AC noise is getting through to my amp. Will need a power conditioner I think.
You do not have the required permissions to view the files attached to this post.
-
grooveguy
- Posts: 467
- Joined: Thu Jun 22, 2006 5:49 pm
- Location: Brea, California (a few miles from Disneyland)
- Contact:
Re: KNOP lathe DIY build
Beautiful renderings, Bryan; reminds me of the "artist's conception" of futuristic cars and space vehicles that once adorned the front cover of Popular Science magazine. Really well done, and sure hope it performs well.
Latency isn't much of an issue in the products we make that use the Sigma part. Actually, we build 'extra latency' into our audio processors to effect what we call "lookahead limiting," an impressive way of saying that the level of the audio can be calculated and gain reduction introduced before the offending peak comes along. Makes for far less clipping, as a fast traditional limiter needs to clip at least the first half-cycle of the waveform before the gain can be reduced. We've recently found that less of the lookahead is required than we originally supposed; a couple of milliseconds is all you need to avoid audible clipping and transient intermod products. Aside from that, the rest of our processing blocks each requires only a sample or two at 48kHz to do their work. Even with the lookahead limiting in the program chain, jocks are still able to monitor themselves off-air with only a hint of 'hollowness' to their voices as the delayed audio reaches their eardrums slightly behind the bone-conducted sound.
Our 'flagship' processor: http://inovonicsbroadcast.com/product/719 uses two Sigma chips, one running at 48kHz to do all the audio processing, and the second runs at 192kHz and generates the FM composite multiplex waveform. Oneof our off-air rebroadcast 'translator' receivers has a single Sigma chip split and running at two different sampling rates. Ah, the wonders of science.
Latency isn't much of an issue in the products we make that use the Sigma part. Actually, we build 'extra latency' into our audio processors to effect what we call "lookahead limiting," an impressive way of saying that the level of the audio can be calculated and gain reduction introduced before the offending peak comes along. Makes for far less clipping, as a fast traditional limiter needs to clip at least the first half-cycle of the waveform before the gain can be reduced. We've recently found that less of the lookahead is required than we originally supposed; a couple of milliseconds is all you need to avoid audible clipping and transient intermod products. Aside from that, the rest of our processing blocks each requires only a sample or two at 48kHz to do their work. Even with the lookahead limiting in the program chain, jocks are still able to monitor themselves off-air with only a hint of 'hollowness' to their voices as the delayed audio reaches their eardrums slightly behind the bone-conducted sound.
Our 'flagship' processor: http://inovonicsbroadcast.com/product/719 uses two Sigma chips, one running at 48kHz to do all the audio processing, and the second runs at 192kHz and generates the FM composite multiplex waveform. Oneof our off-air rebroadcast 'translator' receivers has a single Sigma chip split and running at two different sampling rates. Ah, the wonders of science.
Re: KNOP lathe DIY build
Due to the fact I broke my 2 last needles I had a bit time to work on the cutterhead.
So everything fits,
V spring assembly + torque tube + set screw + stylus is down to 1,92gr
Driver coils + driver surround and cone + pushrod assembly is down to 0,95gr per driver.
main resonance seems to be about 1400Hz, but did not start tests cuts yet. I hope I find some time soon.
So everything fits,
V spring assembly + torque tube + set screw + stylus is down to 1,92gr
Driver coils + driver surround and cone + pushrod assembly is down to 0,95gr per driver.
main resonance seems to be about 1400Hz, but did not start tests cuts yet. I hope I find some time soon.
You do not have the required permissions to view the files attached to this post.
-
grooveguy
- Posts: 467
- Joined: Thu Jun 22, 2006 5:49 pm
- Location: Brea, California (a few miles from Disneyland)
- Contact:
Re: KNOP lathe DIY build
Stunning, Bryan; let's hear it!
Re: KNOP lathe DIY build
Got the cutterhead installed and did some first test cuts. Still got a lot of resonance peaks.
Anyone who can tell me what the dip and peak at 4Khz could be???
But when EQ ed it sounds ok, here's a clip, the needle is chipped hence a lot of bg noise.
https://clyp.it/11bf2mco
Anyone who can tell me what the dip and peak at 4Khz could be???
But when EQ ed it sounds ok, here's a clip, the needle is chipped hence a lot of bg noise.
https://clyp.it/11bf2mco
You do not have the required permissions to view the files attached to this post.
-
grooveguy
- Posts: 467
- Joined: Thu Jun 22, 2006 5:49 pm
- Location: Brea, California (a few miles from Disneyland)
- Contact:
Re: KNOP lathe DIY build
Now, that sounds amazing to me; I caon't catch a hint of that notch, which ought to really be pronounced in a cut like that. All I really can say is, "Wow!"
Re: KNOP lathe DIY build
Hi,
The clip sounds good. Can you post a link to a wav file of the original and the captured playback? I didn't see any way to download the clip from the link you provided.
Do you have any damping material on your 3d printed pusher? That node/anti node peak in the response might be due to the pusher members. If you drop the pusher on a hard surface, does it ring like a bell or thud? If it rings, maybe capture the ringing with a microphone and see if it matches your 4K resonance. Perhaps dipping in plasti-dip or some other material might tame that a bit.
Mark
The clip sounds good. Can you post a link to a wav file of the original and the captured playback? I didn't see any way to download the clip from the link you provided.
Do you have any damping material on your 3d printed pusher? That node/anti node peak in the response might be due to the pusher members. If you drop the pusher on a hard surface, does it ring like a bell or thud? If it rings, maybe capture the ringing with a microphone and see if it matches your 4K resonance. Perhaps dipping in plasti-dip or some other material might tame that a bit.
Mark
Re: KNOP lathe DIY build
http://www.knoprec.be/War%20-%20Low%20Rider.flac
http://www.knoprec.be/War%20-%20Low%20rider%20cut.wav
Just noticed the stereo is reversed haha:)
I will look into the damping issue
http://www.knoprec.be/War%20-%20Low%20rider%20cut.wav
Just noticed the stereo is reversed haha:)
I will look into the damping issue
Re: KNOP lathe DIY build
Hi,
Thanks for posting the two files. I just compared the two using Har-Bal, and I don't see any sign of the resonances. So either you did a great job of EQ'ing them out or they were not really there (see JPG below). Overall, the EQ match looks very good (nice job!). You are a bit bass shy as compared with the original, but that could be easily fixed with some additional EQ. How did you do your EQ? In the screen shot, the cut is in green and the source is red.
Thanks for posting the two files. I just compared the two using Har-Bal, and I don't see any sign of the resonances. So either you did a great job of EQ'ing them out or they were not really there (see JPG below). Overall, the EQ match looks very good (nice job!). You are a bit bass shy as compared with the original, but that could be easily fixed with some additional EQ. How did you do your EQ? In the screen shot, the cut is in green and the source is red.
You do not have the required permissions to view the files attached to this post.
Re: KNOP lathe DIY build
The eq is just really the fabfilter eq match function recorded white noise vs true white noise, a hpf at 17khz and lpf at 30
Re: KNOP lathe DIY build
Hi,
Fab did a really nice job. I should pick Fab up as Har-Bal is getting long in the tooth and Fab has some really nice features.
Not sure why the low end match was a bit off. Was Fab Filter used to match the noise cut and then applied to the War source material? Based on those results, you may not have to go further. But it might be nice to try and tame the source of the secondary resonances.
Mark
Fab did a really nice job. I should pick Fab up as Har-Bal is getting long in the tooth and Fab has some really nice features.
Not sure why the low end match was a bit off. Was Fab Filter used to match the noise cut and then applied to the War source material? Based on those results, you may not have to go further. But it might be nice to try and tame the source of the secondary resonances.
Mark
Re: KNOP lathe DIY build
The EQ match was done using the white noise, afterwards I just load the song in the DAW with the white noise settings in fabfilter. I'm not really shure why it turn out bass shy, but I did got the feeling that in previous heads and EQ attempts fabfilter tended to do the same thing, but not always.
I'm going to test the head as is for a while, and order the parts for a second cutterhead, and experiment on that one. I want to see how far I can push the non feedback design. At the moment i'm just happy the new geometry works out fine.
However i'm not really bothered by technical specs and stuff, I just want to make a nice sounding head which is affordable, durable, is easy to use and build/repair. And cut awesome records, ofcourse:)
I'm going to test the head as is for a while, and order the parts for a second cutterhead, and experiment on that one. I want to see how far I can push the non feedback design. At the moment i'm just happy the new geometry works out fine.
However i'm not really bothered by technical specs and stuff, I just want to make a nice sounding head which is affordable, durable, is easy to use and build/repair. And cut awesome records, ofcourse:)
Re: KNOP lathe DIY build
Hi,
I think I see the problem. Looks like your spectral plot of the white noise has some low frequency junk present (maybe rumble and/or 60Hz mains pickup). Its not really the true response of the cutting system to the noise. Try cutting the noise at a higher level if you can and see if that pulls you out of the mud. The other option would be to cut the noise source without any iRIAA in place and let FabFilter do the RIAA + the correction needed to flatten the head. This should pull the lows out of the mud, however, you might end up with the same issue on the high end. A hybrid approach would be to cut the white noise with only the 2122 Hz high frequency boost portion of the iRIAA curve and let FabFilter deal with the low frequencies. That way you get the best signal to noise ratio during your cut. You would need to process source material using only the iRIAA boost if using this method since FabFilter would be supplying the low frequency portion of the curve.
Hope that makes sense.
Mark
I think I see the problem. Looks like your spectral plot of the white noise has some low frequency junk present (maybe rumble and/or 60Hz mains pickup). Its not really the true response of the cutting system to the noise. Try cutting the noise at a higher level if you can and see if that pulls you out of the mud. The other option would be to cut the noise source without any iRIAA in place and let FabFilter do the RIAA + the correction needed to flatten the head. This should pull the lows out of the mud, however, you might end up with the same issue on the high end. A hybrid approach would be to cut the white noise with only the 2122 Hz high frequency boost portion of the iRIAA curve and let FabFilter deal with the low frequencies. That way you get the best signal to noise ratio during your cut. You would need to process source material using only the iRIAA boost if using this method since FabFilter would be supplying the low frequency portion of the curve.
Hope that makes sense.
Mark
Re: KNOP lathe DIY build
Different monster since I've only been embossing with my head, but I have been doing what Mark suggests - cutting white noise with no iRIAA, putting RIAA on the cut noise and letting fab filter auto EQ - I'm happy with it, but have found the same as you are experiencing - problems with Fabfilter trying to EQ out rumble.
My next step will be to try and subtract the noise floor with RealFir which comes with Reaper, and has a subtract feature which will automatically build a noise profile, from an empty groove.
I've been following your build since before I even started mine and it's great to see and hear such a good result!
My next step will be to try and subtract the noise floor with RealFir which comes with Reaper, and has a subtract feature which will automatically build a noise profile, from an empty groove.
I've been following your build since before I even started mine and it's great to see and hear such a good result!
Record Lathe Embossing Supplies + Embossed Records - http://www.johnnyelectric.co.nz/
Re: KNOP lathe DIY build
Hi,
Great information guys, and amazing progress!
Congratulations Wim!
I'm truly jealous from your results!
Don't worry about the bass cutoff.. I'm sure you ll find a way...
I will be back with my "brother singer" experiments, soon.
Sifis
Great information guys, and amazing progress!
Congratulations Wim!
I'm truly jealous from your results!
Don't worry about the bass cutoff.. I'm sure you ll find a way...
I will be back with my "brother singer" experiments, soon.
Sifis