Inner-workings of the Roland GI-20 and/or the Axon AX50

Started by Kebert_Xela, November 12, 2013, 11:51:13 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Kebert_Xela

Hi, I'm new to this forum, and it seems to me that this community has the most knowledgeable people by-far in regards to guitar-synthesis and MIDI-guitar, so I think I'm in the right place. :)  Hopefully someone might be able to help me.

I was wondering if anyone was familiar with how the Axon Ax50 and/or the Roland GI-20 worked from a circuit/schematic point of view?

I am hoping to design/build something similar to these so that I can use my guitar as a MIDI controller (but I'm not really interested in any on-board synthesis capabilities), so I figured that having a peak inside of either of these would be a good starting point.

If anyone has some amount of working-knowledge of either of these devices, or has any sort of schematics of them, that would be awesome.  :D

Thanks to anyone who can be of help.

admin

#1
Start by researching similar gear

QuoteAndras SzalayOct 23, 2013

I spent most of the last thirty years designing guitar synthesizers. I developed several generations of devices from the Shadow GTM-6 / SH-075 through the Blue Chip Music / Terratec Axon until the Fishman Triple Play today. Over this period I answered thousands of questions from users who had some difficulties to understand certain features or behavior of their devices. Now, as I have here a web page of my company, Panda-Audio, I want to take the opportunity to summarize this information in the form of an FAQ. I hope I can help users of guitar synthesizers a better understanding of their devices with this information. While these answers are based on my designs, I believe most of the information here is true for Roland or other devices too. I will update this page time to time with questions whatever I recall from my memories, or whatever new questions arise.
See that on my web page:

http://www.panda-audio.com/midiguitar.php


http://www.synfo.nl/servicemanuals/Roland/GR-30_SERVICE_NOTES.pdf

http://www.joness.com/gr300/service.htm

http://www.joness.com/gr300/service/GR-700_SERVICE_NOTES.pdf

and Google search on "Guitar to MIDI Patents" yields:

http://www.google.com/patents/US5033353


This one is rather detailed
http://www.faqs.org/patents/app/20100242712

Kebert_Xela

Hey, thanks.  :D  This definitely leads me in the right direction.

According to what I have come across on some other discussions on this forum and elsewhere, it seems that the GR-300 is considered the benchmark in performance for pitch-detection (which is probably why it was one of the devices you referred me to). I was actually able to find a pdf of the patent for it as well.

Considering that the GR-300 was made in the 1980's, I wonder why pretty much all relatively newer devices do not perform with comparable levels of latency minimization.  ???  I'm sure the conversion to MIDI messages something to do with it, but even some of the analog devices made later still were not as good.

I would ask why, but I feel this is probably a question to which there is really no simple answer, and I would imagine that almost everyone here probably wonders about it too. :(

montyrivers

Quote from: Kebert_Xela on November 17, 2013, 05:51:30 PM
Hey, thanks.  :D  This definitely leads me in the right direction.

According to what I have come across on some other discussions on this forum and elsewhere, it seems that the GR-300 is considered the benchmark in performance for pitch-detection (which is probably why it was one of the devices you referred me to). I was actually able to find a pdf of the patent for it as well.

Considering that the GR-300 was made in the 1980's, I wonder why pretty much all relatively newer devices do not perform with comparable levels of latency minimization.  ???  I'm sure the conversion to MIDI messages something to do with it, but even some of the analog devices made later still were not as good.

I would ask why, but I feel this is probably a question to which there is really no simple answer, and I would imagine that almost everyone here probably wonders about it too. :(

  The same reason playing an ARP synthesizer with a midi to voltage converter produces different results than an old school voltage sequencer.  Midi is digital so it only speaks off and on.  Note on/off plus velocity.  There are a lot of controllers and converters that try to mimic note modulation by sending additional control data (keyboards with after touch) but nothing will ever compare to the response and expressiveness of the genuine article.

This is why many guitar synth enthusiasts on this forum are loathe to use midi to power a synth in the first place, they just go for a modular device, vg or gr 300

Kebert_Xela

#4
Quote from: montyrivers on November 17, 2013, 07:45:12 PM
  The same reason playing an ARP synthesizer with a midi to voltage converter produces different results than an old school voltage sequencer.  Midi is digital so it only speaks off and on.  Note on/off plus velocity.  There are a lot of controllers and converters that try to mimic note modulation by sending additional control data (keyboards with after touch) but nothing will ever compare to the response and expressiveness of the genuine article.

Thank you montyrivers for your input. Yes, as I have discovered, Midi has the limitations of not being able to capture much of the more expressive aspects of guitar playing, and these would have to be added with external controls to adjust the parameters of pitch bend, modulation, etc. Unfortunately, it seems that this cannot be remedied in a way that might hope to satisfy more people until a more involved protocol is developed as a successor to Midi (which is supposedly in the works, but that could mean in 10 years - see here http://www.midi.org/aboutus/news/hd.php).

I was more concerned with why the purely pitch-tracking aspect has not been what been as successful in the digital devices as it is in the analog ones.

http://www.joness.com/gr300/MIDI_SPEED.htm

From my understanding of this, in addition to what I have seen and heard in various places, including here (and correct me if I'm wrong), the purely signal processing portion of the GR-300 provided the fastest and most precise pitch-tracking scheme that was ever implemented.

That being said, from my understanding of the pitch-detection circuit in the GR-300 (and again, anyone feel free to correct me if I'm wrong or lacking a full understanding), the fundamental frequency of each string is isolated by a filter (what they call the "variable bandpass filter") and locked in a type of feedback control circuit so as to prevent the other harmonics from sounding after the attenuation of the fundamental. Then, I believe they detect the zero-crossings to determine the frequency, which is output by a square-wave signal at this portion of the signal processing design.

It is what happens after this point where I am a little confused.

  • Partly, from the remaining portion of the processing circuit in the GR-300, which involves creating ramp functions (a sawtooth signal) from very short pulses obtained from the square-wave signal, and then limiting the magnitude of the amplitudes. I am a little perplexed as to the purpose this serves. I know that the sawtooth wave is often the choice of signal in driving analog synthesizers, but I think I am still missing something that would
  • Partly because, at this point, from a circuit analysis point of view (from looking at the schematic in the Service notes, it's around the T/V and V/T part in the upper right section), various transistors and other components have their operational characteristics being determined from voltage levels from various other parts elsewhere in the device. I know that this is fairly common in circuits, especially those with lots of transistors, but it doesn't make it any less overwhelming for me, being that I am relatively new to this.


In a roundabout way, I guess what I am wondering is why the fundamental-detection and pitch-tracking design of the GR-300 was never taken and implemented for conversion to digital signals (and then ultimately into Midi messages)?

Is it because of the demands of processing speeds for the digital information? Speeds that perhaps may be achievable now, but were not when digital and Midi guitar synthesis was first being designed, either due to budgetary reasons or due to the scope of what was technologically possible at the time?

Or is there something that I am overlooking or misunderstanding?

Kebert_Xela

...
...
.. :o

Realized how much of a read that is.  :-[

MUCH appreciated for anyone that can bear with me through that.
Hopefully, what it lacks in succinctness, it makes up for in an otherwise better understanding of my thought process and what might be escaping me.

Elantric

#6
I see Roland as being a major player in the Guitar Synth evolution 
http://www.roland.com/V-Guitar/


IMHO - it boils down to MIDI - which is a early 1980's serial communications protocol shackled at 31.25kHz Baud Rate.

Guitar synthesis became hampered as soon as they tried to to support playing MIDI Keyboard with a Guitar.

observe the 1980 GR-300 has no A/D / DSP / D/A hardware  - no MIDI !

IMHO  - compared to todays hardware, the GR-300 is more a mix of analog and early Don Lancaster CMOS Cookbook type circuits.   


True you were limited to a "GR-300" tones, but they reacted fast with almost zero latency.
http://www.joness.com/gr300/MIDI_SPEED.htm
http://www.pfeiffermike.com/2011/03/electronics-books.html



It was not until the mid 1990s Roland "VG-x" COSM systems that fast real time low latency Guitar Synthesis / Modeling started happening again - for mere mortals!
Its a sad historical fact Roland was unable to communicate to the masses their ""VG-x" COSM system" (perhaps because a VG-8 was $2500 back in 1996) and left it to others (Line-6) to get all the credit for this technology( Variax / POD, etc)

Kebert_Xela

So for the first part of what you are saying, do you mean

  • that the rate of MIDI messages that need to be sent is too slow given MIDI's baud rate (which can handle about 180 or so 3-byte messages every second)?
or
  • that the MIDI baud rate puts hampering limitations on the frame of time between when the pitch-tracking signals are processed and when the corresponding MIDI messages are output?
I guess another way of putting it would be:

Are the delay issues a result of the size and quantity of bit-information needed in the MIDI messages exceeding the transmission rate?

Or is it that the rate of transmission allows for only a very short window of time for which everything in between the signals being output from the pitch-detection/pitch-tracking circuits are processed and the forming of the MIDI messages can occur?




Also, I was a little confused on the last part about the Line-6 PODs. Were simply saying that they are the decendants of Roland's COSM system, but since they were made years later, they could more cheaply handle the processing demands of the analog signals (and thus get all the credit)?

I thought the PODs were basically just a whole system of analog effects in a small package and did not have any Midi-conversion capabilities.

Elantric

#8
QuoteI thought the PODs were basically just a whole system of analog effects in a small package

All Line6 gear is based upon this topology

A/D> DSP >D/A

This includes all Pods, M13/M9/M5, ToneCore pedals, Variax, Tyler Variax, etc.

The power is in the dsp algorithms
http://www.dspalgorithms.com/

Roland pioneered all this with the 1995 VG-8 - i.e. Guitar PU modeling, Alt tunings, Tube Amp simulation with added element of polyphonic HRM Synths  - something Line 6 stops short of implementing since Line6 separates the hexaphonic 6 channel Guitar Modeling DSP into the guitar then only sends a 2 channel digital link (VDI cable) to the floor based Amp modeling DSP (POD HD500) 


Quoteand [I thought Line6] did not have any Midi-conversion capabilities.
True - but they do have fast monophonic "Guitar to Synth" tones that can be triggered by any Guitar in the POD HD-500  / HD -Pro 


If its the best Guitar to MIDI interface you seek in 2013 - then look no further than a Fishman Tripleplay - created by the Design Engineer of the Axon Guitar to MIDI System  = Andras Szalay
Fishman TriplePlay TOP THINGS TO KNOW
https://www.vguitarforums.com/smf/index.php?topic=8427.msg60207#msg60207


Andras Szalay wrote>
Quote
I spent most of the last thirty years designing guitar synthesizers. I developed several generations of devices from the Shadow GTM-6 / SH-075 through the Blue Chip Music / Terratec Axon until the Fishman Triple Play today. Over this period I answered thousands of questions from users who had some difficulties to understand certain features or behavior of their devices. Now, as I have here a web page of my company, Panda-Audio, I want to take the opportunity to summarize this information in the form of an FAQ. I hope I can help users of guitar synthesizers a better understanding of their devices with this information. While these answers are based on my designs, I believe most of the information here is true for Roland or other devices too. I will update this page time to time with questions whatever I recall from my memories, or whatever new questions arise.


http://www.panda-audio.com/midiguitar.php

http://www.panda-audio.com/designs.php