Wireless GK

Basics

The Roland GR synthesizer and its VG-8 companion are incredible instruments. The dumbness of the 13-pin Atari monitor plug is incredible too. Either you inadvertently rip out your cable from the GK controller, or the cable wraps itself around a foot pedal, or it starts to twist all along. It's high time for a wireless GK cable. This is what this proposal is about.

The 7 audio signals (guitar, GK string 1-6), synth vol and S1/S2 are FM modulated and fed into a cheap commercial video link in the 2.4 or 5.8 GHz range (ISM, license free in most countries). The DIY project stops at the video signal output. The DIY'er has to provide his or her own video transmitter link i.a.w. national regulations, as well as the power supply.

Expected performance:

range about 300 m free air, 30 inside buildings.
S/N = 90 dB

The video link will perform best in an environment without much noise (what else is new?). Everyone transmits on 2.4 GHz now. Diversity receiver preferred, possibly in the 5.8 GHz ISM band. Extremely expensive, it seems.

The downsode of this analog solution is many potentiometers to tune...

Transmitter

Proposed block diagram see here.

To enhance the overall S/N ratio by up to 45 dB, the 6 GK signals and the guitar signal are compressed by hi-fi grade compressors using the popular chip SA571. Application circuits see Philips Semiconductor AN174. The compressor application needs one audio OP plus 2/4 LM324 per channel so we end up needing 3x LM833/NE5532/TL072 plus one NE5534/TL071 plus 3x LM324 plus 1x LM4558. Compressor distortion must be trimmed.

Then each signal is applied to a ultra-linear VCO (XR2209). Center frequencies: 300 ... 900 kHz, sweep = +/- 30 kHz (although the XR215A PLL data sheet says you need to limit FM deviation to 1% to get a low THD response). . The 7 FM triangle-wave signals are lowpassed (7/4 quad OPs) to keep interchannel crosstalk low, then added (1/4 quad OP) and attenuated to 1Vpp, then fed into the video transmitter. The TL074 OPs have a GBW of 4 MHz, and they are operated with unity gain so that we don't require more expensive video OPs. Makes a nice 4 MHz low pass as a side effect too. Stability should be granted up to about 1 MHz. Alternatives: National LM6152 or CLC414 or Burr-Brown OPA4134, or Burr-Brown OPA4350) (pin compatible).

Synth-Vol is VCO converted to f₀ = 100 kHz +/- 10% (fmin=80 kHz, fmax=120 kHz) using a cheap CD4046 VCO/PLL. The 4046 is supplied with +/- 7V. The center frequency is defined as 0V. S1 pulls the VCO input to -7V, and S2 pulls it to +7V. The square wave output is low pass filtered. The resulting signal is added to the 7 FM signals above so that we have one signal with a bandwidth of about 1MHz.

Power Supply - Estimated power requirement:

7 * XR2209	35 mA
4 * NE571	14 mA
1 * CD4046 @ 100 kHz	0,5 mA
4 * TL074	22 mA
MAX665 & MAX761	only several µA quiescent current
Sum:	~ 70-75 mA

if using a 12V transmitter: it is recommended to use a 7.2V Li-Ion or NiMH battery pack with about 7-15 Wh. The raw battery voltage needs not to be filtered and can be used directly. -7V is generated by a DC-DC converter MAX665 (45 kHz). +12V for the transmitter are generated by a low-power step-up converter MAX761. MAX761 guarantees 150mA at an input voltage of 5V. That should really be sufficient.
if using a 5V transmitter: use 4 AA type NiMH cells. The largest capacity I've seen so far is 1700mAh. Step Up converter MAX761 generates +7V. Inverter MAX665 generates -7V @ 45 kHz. Transmitter is powered by the raw battery voltage. Instead of the NiMH cells you can also use Rayovac or Accucell rechargable batteries. Or even normal Alkalines if your little heart desires. Actually, this solution is much easier than the 7.2V Li-Ion stuff because mechanically it is much harder to make a compartment and contacts for the Li-Ion pack than for 4 AA cells. I tend to the NiMH solution ... With the small ELV transmitter ALM 2441/2442, 4 1500mAh cells last about 4.5 hours.

The options are selectable by 4 jumpers. +7V and -7V are also fed to the GK controller.

Most components should be SMD. The XR2209 VCO isn't available in a SMD package, though :-( Everything should fit on a 80x100mm PCB (half Euro card, that's what the freeware Eagle layouter can provide), maybe a sandwiched 80x70mm daughter card will be needed. Since the 13-pin PCB connector is 22 mm high including leads the case should be 24-25 mm high internally at least anyway. An alternative would be to use no 13-pin PCB connector (who makes the 30cm 13-pin stub cable?) but permanently attach the cable to the PCB. The PCB needs to provide a 14-pin in-line plug too.

Remark: I initially intended to use the audio channel of the video link for the synth vol FM converted frequency. But people may want to use video transmitters without audio channel. If you have a transmitter with audio channel you can perhaps use it for a headset microphone for singing... this is where a diversity receiver comes in handy... Or should we use the free half of the NE571 for a microphone channel, at the expense of one more VCO/PLL pair plus a mic transformer? Mike Murphy pointed out that NTE still has replacement parts for the discontinued NE565/566 combo.

Receiver

Proposed block diagram see here.

The video signal is fed to 8 narrow band pass filters.

The 7 upper outputs are fed into 7 ultra-linear PLL circuits (XR215A). The demodulated outputs are expanded by NE571's, and gain corrected in the output. You will need to adjust the GK pickup sensitivity accordingly. Performance: the XR215A data sheet talks about typically 65 dB S/N, and the NE571 data sheet talks about up to 45 dB better S/N, but has 90 dB S/N itself so this is probably what we will reach.

The lower frequency (synth vol & S1/S2) is schmitt-triggered and limited to +/- 3V, then FM demodulated by a CD4046 and fed into a window discriminator. If the voltage is below 0V, S1 is detected, and the Synth Volume is pulled to 0V (Mute). If the voltage is above +5V, S2 is detected, and the Synth Volume is pulled to 0V (Mute). If the voltage is between 0V and +5V, Mute is not asserted. The demodulated output source follower's offset must be observed, or a FET input buffer OP like TL071 must be used to buffer directly from the PLL low pass filter. Since we need to buffer anyway, the OP is preferred. It allows us also to re-scale the synth vol voltage if required.

The receiver PLLs show their lock status with LEDs (not XR215A)

Power supply: small wall mart, +/- 7V or something.

3 selectable GK output sockets, Guitar Out,

True Diversity Receiver Idea

Someone suggested to do diversity on the demodulated end of the receiver(s). This would require

2 or 3 low-cost receivers of the type above, all running on the same 2.4 GHz channel.
2 or 3 video OPs with enable (e.g. OPA3355). Their common output is routed to the demodulation circuit described above. Soft switching would make sense to avoid clicks. Maybe even a 74HC4066 is sufficient (max on-channel bandwidth 160 MHz, crosstalk -80 dB @ 1 MHz, off-channel isolation -40 dB @ 1 MHz).
2 or 3 full wave rectifiers to measure the video signal strength at the demod outputs. Since the rectified signal has a bandwidth of about 1 MHz the low pass time constant can be in the low millisecond range.
A 5..10 MHz bandpass per receiver to detect noise whenever the FM signal strength is too low
a microcontroller (e.g. PIC with AD converters) to choose the strongest receiver after a certain dead time and with a floating hysteresis. A 20 MHz PIC (e.g. 16F85/86) is fast enough to provide a detection cycle time in the millisecond range. Algorithm suggestions welcome!
If we limit ourselves to using 2 receivers (what most commercial systems do) the detector / switch part can be replaced by hardware (2 field strength detectors, Schmitt Trigger with floating hysteresis and delay, 2-channel video OP w/ enable (e.g. OPA2680).

Or does someone have a working application for the Analog Devices AD6600?

This is independant of the other stuff and can be built in addition. More design ideas are here.

http://www.dynapix.com/supertriad.htm ?

A very interesting system is S-515P on http://www.cocom.com/acatalog/Online_Catalog_CVT_500_29.html

Video TX/RX sets

    - Conrad pp. 519/882. 
    - the Gigalink module offers 5 selectable channels. 
    -> use alarm switch connector for S1/S2 ?
    -> use microphone input for synth vol VCO signal
    -> hence only 7 audio channels over video FM. 

    ELV (www.elv.de): 4-Kanal-Sendesystem ALM 2442 399 DM (set!)
    - transmitter uses 4.8..7.2V, i.e. Vbat directly. 
    - can use a case w/ 4 AA-cell battery compartment. 
    - batt cptmt: Conrad 521868 or external battery
      pack w/ cable
    - also nice: Bopla BOS 752 or 807
      (http://www.bopla.de/, w/ 4 AA-cell
      battery compartment) (Reichelt pg 179/80) 


    https://spynet.xynx.com:
    - auch 25 mW Minisender.  (auch Gigalink - 2004311, 390 DM)
    - angeblich nur 120 mA 
    - Mini TX II 10 oder 25 mW (?) 2200012  480 DM
    - Empfänger 2048020 229 DM (COnrad 329 DM) 

    -> non-diversity :-( 

    http://www.securitec-gerlach.de/: very
      cheap. Quality? 

    http://www.dynapix.com/oem.htm ? Only signel
      frequency, needs 500 mA!

    Diversity: 
    http://www.premierwirelessinc.com  5.8 GHz

The old proposal

I was asked why I dropped my old proposal which involved a digital transmission. Well you simply can't get 5 MBit/s through a 5 MHz channel, whatever coding you choose. MP3 isn't an option either - it produces latencies of 50 ms and more. One more reason not to pursue this idea is the logistics - how do DIYers get pre-programmed PICs? Limiting the idea to people who can program PICs isn't really an option.

To my other electronic projects

Harald Milz

Last modified: Tue May 15 11:33:13 CEST 2001