REVIEW: Ploytec GM5 USB Midi Interface

[Altitude909]

This chip was brought to my attention via the Midibox forums and is produced by a company called Ploytec (Ploytec GmbH . USB Audio and more) out of Germany who produces various midi odds and ends. They also offer the PCB for this project but I opted for the Midibox (tk_gm5_bulkorder [MidiBox.org]) version which has the correct jumper pinout for connecting to a Core module. The uC itself is an Atmel AVR AT90USB162 which is a very high performance 8 bit chip with 16kb of FLASH and a onboard 512b EEPROM. The price of the chip is 4.50 € and the 1x1 pcb is another 3.00 € ($ 10.50 US). Parts from Mouser came to another $12.00. It is worth mentioning that this chip is a TQFP32 package which is a SMD so you will need an appropriate iron to solder this in.

The chip is class compliant so it runs without and drivers under win32/win64/mac/linux but also has a dedicated driver for win32 to overcome the shortcomings of the generic windows midi driver. 5 in/out ports are available per chip and a 5x5 PCB is in the works but for my tests I am only using the 1x1 PCB which was the only thing available at that time.

Test Setup: Vista32, same USB port, same midi cable with a loop back using Miditest 4.6. I run 2 Midiex 8s and also had some other interfaces (my Novation controller and a Midisport 1x1) to test against. For those who don't know how to interpret these results, low numbers and tight groupings are what you want to see.

To summarize that rather long list, the GM5 bests the Midex 8 by about a factor of 2 pretty much down the board and by about a factor of 3 for the midisport. The max deviation is also very impressive with max of 1.67 (!) ms making this the tightest timed midi interface I have used to date. For less than $25, I really dont think it gets any better than this..

The GM5:

--------------------------------------------------------------------------
MidiTest Results
--------------------------------------------------------------------------



================ Info ====================================================

Date: 19 Dec 2008
Time: 07:16:20
AppVersion: 4.6.231
OS: Professional, Service Pack 1 (Build 6001)
Processor(s): Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
Speed: 2400 MHz
Number: 4
API: MultiMedia Extensions (MME)
Test type: Advanced
Use timestamp: yes
Errors: None



================ Tested Message Types ====================================

Note off: yes
Note on: yes
Key aftertouch: yes
Controller: yes
Program change: yes
Channel aftertouch: yes
Pitchbend: yes
System exclusive: yes
MIDI time code quarter frame: yes
Song position pointer: yes
Song select: yes
Tune request: yes
MIDI clock: yes
MIDI tick: no
Start: yes
Continue: yes
Stop: yes
Active sensing: yes
System reset: yes
System exclusive mixed with realtime messages: no



================ Ports ===================================================

MIDI Output: GM5 port 1
Description: midibox.org GM5
Provider: midibox.org
DriverDate: 11-20-2008
DriverVersion: 1.0.6.0


MIDI Input: GM5 port 1
Description: midibox.org GM5
Provider: midibox.org
DriverDate: 11-20-2008
DriverVersion: 1.0.6.0



================ Results Per Message =====================================

MESSAGES Snd Rcv Snd+Rcv

Message TotalTime: 2026.36 ms 10170.38 ms 12196.74 ms
Message MaximumTime: 0.18 ms 1.67 ms 1.73 ms
Message MinimumTime: 0.03 ms 0.00 ms 0.04 ms
Message AverageTime: 0.06 ms 0.33 ms 0.39 ms
SysexTime: 22.56 ms 2473.62 ms 2496.18 ms
SysexAverage: 0.00 ms 0.25 ms 0.25 ms

< 1 ms: 31250 28461 27792
1 - 2 ms: 0 2789 3458
2 - 3 ms: 0 0 0
3 - 4 ms: 0 0 0
4 - 5 ms: 0 0 0
5 - 10 ms: 0 0 0
10 - 20 ms: 0 0 0
20 - 50 ms: 0 0 0
50 - 100 ms: 0 0 0
> 100 ms: 0 0 0

Message count: 31250 Sysex count: 155
Sysex size: 9999 Sysex passed: 9999

Message latency: 0.39 ms Total time: 64.520 sec
Message jitter: 0.41 ms
Message max deviation: 1.34 ms



================ Results Per Byte ========================================

BYTES

Byte TotalTime: 4862.94 ms
Byte MaximumTime: 1.18 ms
Byte MinimumTime: 0.01 ms
Byte AverageTime: 0.16 ms

< 1 ms: 31249
1 - 2 ms: 1
2 - 3 ms: 0
3 - 4 ms: 0
4 - 5 ms: 0
5 - 10 ms: 0
10 - 20 ms: 0
20 - 50 ms: 0
50 - 100 ms: 0
> 100 ms: 0

Byte count: 78863

Byte latency: 0.16 ms
Byte jitter: 0.16 ms
Byte max deviation: 1.02 ms

Midex 8 with Directmusic Driver:

--------------------------------------------------------------------------
MidiTest Results
--------------------------------------------------------------------------



================ Info ====================================================

Date: 19 Dec 2008
Time: 07:09:37
AppVersion: 4.6.231
OS: Professional, Service Pack 1 (Build 6001)
Processor(s): Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
Speed: 2400 MHz
Number: 4
API: DirectMusic
Test type: Advanced
Use timestamp: yes
Errors: None



================ Tested Message Types ====================================

Note off: yes
Note on: yes
Key aftertouch: yes
Controller: yes
Program change: yes
Channel aftertouch: yes
Pitchbend: yes
System exclusive: yes
MIDI time code quarter frame: yes
Song position pointer: yes
Song select: yes
Tune request: yes
MIDI clock: yes
MIDI tick: no
Start: yes
Continue: yes
Stop: yes
Active sensing: yes
System reset: yes
System exclusive mixed with realtime messages: no



================ Ports ===================================================

MIDI Output: Midex8 8
Description: Steinberg Midex 8 USB Midi Driver
Provider: Steinberg
DriverDate: Not available
DriverVersion: 1.9.0.3


MIDI Input: Midex8 8
Description: Steinberg Midex 8 USB Midi Driver
Provider: Steinberg
DriverDate: Not available
DriverVersion: 1.9.0.3



================ Results Per Message =====================================

MESSAGES Snd Rcv Snd+Rcv

Message TotalTime: 2698.02 ms 21399.95 ms 24097.97 ms
Message MaximumTime: 1.53 ms 4.17 ms 4.26 ms
Message MinimumTime: 0.05 ms 0.00 ms 0.05 ms
Message AverageTime: 0.09 ms 0.68 ms 0.77 ms
SysexTime: 27.38 ms -375.34 ms -347.97 ms
SysexAverage: 0.00 ms -0.04 ms -0.03 ms

< 1 ms: 31249 25264 24461
1 - 2 ms: 1 2570 3210
2 - 3 ms: 0 2080 1985
3 - 4 ms: 0 1323 1566
4 - 5 ms: 0 13 28
5 - 10 ms: 0 0 0
10 - 20 ms: 0 0 0
20 - 50 ms: 0 0 0
50 - 100 ms: 0 0 0
> 100 ms: 0 0 0

Message count: 31250 Sysex count: 153
Sysex size: 9999 Sysex passed: 9999

Message latency: 0.77 ms Total time: 347090.001 sec
Message jitter: 0.88 ms
Message max deviation: 3.49 ms



================ Results Per Byte ========================================

BYTES

Byte TotalTime: 9820.41 ms
Byte MaximumTime: 3.47 ms
Byte MinimumTime: 0.02 ms
Byte AverageTime: 0.31 ms

< 1 ms: 28684
1 - 2 ms: 2294
2 - 3 ms: 221
3 - 4 ms: 51
4 - 5 ms: 0
5 - 10 ms: 0
10 - 20 ms: 0
20 - 50 ms: 0
50 - 100 ms: 0
> 100 ms: 0

Byte count: 78795

Byte latency: 0.31 ms
Byte jitter: 0.40 ms
Byte max deviation: 3.16 ms

Midex 8 with MME Driver:

--------------------------------------------------------------------------
MidiTest Results
--------------------------------------------------------------------------



================ Info ====================================================

Date: 19 Dec 2008
Time: 07:13:57
AppVersion: 4.6.231
OS: Professional, Service Pack 1 (Build 6001)
Processor(s): Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
Speed: 2400 MHz
Number: 4
API: MultiMedia Extensions (MME)
Test type: Advanced
Use timestamp: yes
Errors: None



================ Tested Message Types ====================================

Note off: yes
Note on: yes
Key aftertouch: yes
Controller: yes
Program change: yes
Channel aftertouch: yes
Pitchbend: yes
System exclusive: yes
MIDI time code quarter frame: yes
Song position pointer: yes
Song select: yes
Tune request: yes
MIDI clock: yes
MIDI tick: no
Start: yes
Continue: yes
Stop: yes
Active sensing: yes
System reset: yes
System exclusive mixed with realtime messages: no



================ Ports ===================================================

MIDI Output: 2- Midex8 8
Description: Not available
Provider: Not available
DriverDate: Not available
DriverVersion: Not available


MIDI Input: 2- Midex8 8
Description: Not available
Provider: Not available
DriverDate: Not available
DriverVersion: Not available



================ Results Per Message =====================================

MESSAGES Snd Rcv Snd+Rcv

Message TotalTime: 1913.16 ms 29946.21 ms 31859.37 ms
Message MaximumTime: 0.40 ms 4.14 ms 4.20 ms
Message MinimumTime: 0.04 ms 0.00 ms 0.04 ms
Message AverageTime: 0.06 ms 0.96 ms 1.02 ms
SysexTime: 21.78 ms 3269.69 ms 3291.46 ms
SysexAverage: 0.00 ms 0.33 ms 0.33 ms

< 1 ms: 31250 18288 17842
1 - 2 ms: 0 6924 6959
2 - 3 ms: 0 4562 4805
3 - 4 ms: 0 1464 1614
4 - 5 ms: 0 12 30
5 - 10 ms: 0 0 0
10 - 20 ms: 0 0 0
20 - 50 ms: 0 0 0
50 - 100 ms: 0 0 0
> 100 ms: 0 0 0

Message count: 31250 Sysex count: 157
Sysex size: 10001 Sysex passed: 10001

Message latency: 1.02 ms Total time: 147.798 sec
Message jitter: 1.04 ms
Message max deviation: 3.18 ms



================ Results Per Byte ========================================

BYTES

Byte TotalTime: 12895.09 ms
Byte MaximumTime: 3.11 ms
Byte MinimumTime: 0.01 ms
Byte AverageTime: 0.41 ms

< 1 ms: 27910
1 - 2 ms: 3083
2 - 3 ms: 250
3 - 4 ms: 7
4 - 5 ms: 0
5 - 10 ms: 0
10 - 20 ms: 0
20 - 50 ms: 0
50 - 100 ms: 0
> 100 ms: 0

Byte count: 78979

Byte latency: 0.41 ms
Byte jitter: 0.44 ms
Byte max deviation: 2.70 ms

Midi sport 1x1:

--------------------------------------------------------------------------
MidiTest Results
--------------------------------------------------------------------------



================ Info ====================================================

Date: 22 Dec 2008
Time: 18:03:52
AppVersion: 4.6.231
OS: Professional, Service Pack 1 (Build 6001)
Processor(s): Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz
Speed: 2400 MHz
Number: 4
API: MultiMedia Extensions (MME)
Test type: Advanced
Use timestamp: yes
Errors: None



================ Tested Message Types ====================================

Note off: yes
Note on: yes
Key aftertouch: yes
Controller: yes
Program change: yes
Channel aftertouch: yes
Pitchbend: yes
System exclusive: yes
MIDI time code quarter frame: yes
Song position pointer: yes
Song select: yes
Tune request: yes
MIDI clock: yes
MIDI tick: no
Start: yes
Continue: yes
Stop: yes
Active sensing: yes
System reset: yes
System exclusive mixed with realtime messages: no



================ Ports ===================================================

MIDI Output: Out USB Midi 1x1
Description: Not available
Provider: Not available
DriverDate: Not available
DriverVersion: Not available


MIDI Input: In USB Midi 1x1
Description: Not available
Provider: Not available
DriverDate: Not available
DriverVersion: Not available



================ Results Per Message =====================================

MESSAGES Snd Rcv Snd+Rcv

Message TotalTime: 194.48 ms 67939.69 ms 68134.17 ms
Message MaximumTime: 0.03 ms 5.95 ms 5.95 ms
Message MinimumTime: 0.00 ms 0.00 ms 0.00 ms
Message AverageTime: 0.01 ms 2.17 ms 2.18 ms
SysexTime: 15.54 ms 505.73 ms 521.27 ms
SysexAverage: 0.00 ms 0.05 ms 0.05 ms

< 1 ms: 31250 7511 7466
1 - 2 ms: 0 6833 6825
2 - 3 ms: 0 7313 7324
3 - 4 ms: 0 6627 6630
4 - 5 ms: 0 2868 2906
5 - 10 ms: 0 98 99
10 - 20 ms: 0 0 0
20 - 50 ms: 0 0 0
50 - 100 ms: 0 0 0
> 100 ms: 0 0 0

Message count: 31250 Sysex count: 152
Sysex size: 10000 Sysex passed: 10000

Message latency: 2.18 ms Total time: 178.830 sec
Message jitter: 1.32 ms
Message max deviation: 3.77 ms



================ Results Per Byte ========================================

BYTES

Byte TotalTime: 29007.20 ms
Byte MaximumTime: 4.43 ms
Byte MinimumTime: 0.00 ms
Byte AverageTime: 0.93 ms

< 1 ms: 18111
1 - 2 ms: 11136
2 - 3 ms: 1595
3 - 4 ms: 399
4 - 5 ms: 9
5 - 10 ms: 0
10 - 20 ms: 0
20 - 50 ms: 0
50 - 100 ms: 0
> 100 ms: 0

Byte count: 78871

Byte latency: 0.93 ms
Byte jitter: 0.66 ms
Byte max deviation: 3.50 ms

[gl3ny]

The numbers look good with midi test but what about with a host application running, like cubase for instance?

Do you have any results for this?

[Altitude909]

I'll take a look at it tomorrow. What's your test procedure

*edit

Just saw your post in the other thread, send me your spreadsheet and midi file and i'll run the test. Altitude909 at gmail

[TS-12]

I wonder if there would be any noticeable timing problems to the ear when sending about 12 midi tracks to midi device with about 48 notes (4 notes per each track) at 1/6th notes from software sequencer to a midi module or keyboard? can you do a test? I tried many different midi interfaces and can't find one that will not have noticeable timing shifting/delays after having mulpitple midi tracks going to external midi devices.

[Altitude909]

Originally Posted by TS-12

I wonder if there would be any noticeable timing problems to the ear when sending about 12 midi tracks to midi device with about 48 notes (4 notes per each track) at 1/6th notes from software sequencer to a midi module or keyboard? can you do a test? I tried many different midi interfaces and can't find one that will not have noticeable timing shifting/delays after having mulpitple midi tracks going to external midi devices.

Seriously? 12 midi tracks to/through one device? You do understand that midi is a serial protocol so if you have 12 events at the same time in your sequencer, they will be transmitted sequentially, there will always be a time offset from the first and last note and with 12 notes, it will be very noticeable.

[TS-12]

Originally Posted by Altitude909

Seriously? 12 midi tracks to/through one device? You do understand that midi is a serial protocol so if you have 12 events at the same time in your sequencer, they will be transmitted sequentially, there will always be a time offset from the first and last note and with 12 notes, it will be very noticeable.

really? (midi is super tight between hardware sequencer and keyboards connections) So if I have alot of MIDI modules and keyboards and synths, I am better off using harware sequencer?


I had Triton Extreme was using it through its USB-MIDI port. I had about 10 tracks going to it withought very noticeable MIDI sloppiness. Now I have MOTIF-ES (as a main keyboard) its MIDI-USB has the worst MIDI timing I have ever heard.

With all this new technology and Copmuter power today, there still can't be an accurate and fast MIDI between Computer and multitimbral MIDI sound module.

[Altitude909]

Its really more of a midi thing. There is a lot of problems with midi over usb but u also have to take into consideration the limitations of the protocol as well. I deal with it by using a single midi out for each instrument, then I dont have to worry about traffic becoming an issue. Hardware sequencers wont have the slop or latency of usb midi but will still have the same problem of sending too much info

[golden beers]

yea GM5 has good results for USB..!

but still not as good as a $5 soundblaster clone, check Maximum Snd+Rcv.:

Message______________Snd________ Rcv__________Snd+Rcv

Message MaximumTime: 0.13 ms ____0.39 ms______ 0.44 ms
Message MinimumTime: 0.03 ms _____0.27 ms _____0.38 ms
Message AverageTime: 0.03 ms _____0.36 ms _____0.40 ms



but one thing this test does not cover is dropped data.

ie if you try hitting 100 midi notes at the same moment. the USB system (should) have a buffer. don't know about the soundblaster, it would probably drop some notes here.

[Altitude909]

Ha! I kept my SB around for a long time for that exact reason.

[TS-12]

Sound Blaster? a consumer sound card has better MIDI than all these "Pro" ones? Is there a specific Sound Blaster Model or all Creative Sound Blaster Cards have good MIDI?

[tvsky]

can you do the midi test over a full songs with of midi data , about 6 mins worth? pretty sure the default test is only 15 or 30sec

I found the longer you pushed out the test the more you got large max deviation as the computer decided to do something else .

[Altitude909]

Originally Posted by TS-12

Sound Blaster? a consumer sound card has better MIDI than all these "Pro" ones? Is there a specific Sound Blaster Model or all Creative Sound Blaster Cards have good MIDI?

I had a sblive, I think it was the fact that it was pci. Not many sound interfaces made in pci anymore

[golden beers]

Originally Posted by TS-12

Sound Blaster? a consumer sound card has better MIDI than all these "Pro" ones? Is there a specific Sound Blaster Model or all Creative Sound Blaster Cards have good MIDI?

its just the serial (or joystick) port on a PCI card works faster than USB, doesnt matter what card.

I don't actually know why this should be as Universal Serial Bus is also a serial connection but it does NOT work faster from the PCI slot (why would it?): I know because ive tested with a PCI USB card, which performed worse (as you would expect) then the USB slots directly connected to the Mo Bo.

i suspect if you had an old fashioned printer port you would also have a super tight midi system.

it's just easier with soundcards because they have a 'game controller' driver which has had midi bolted on to it for years. it just works.

having said this, i use USB. my stuff is tight enough. it's really about how uniform your midi timing is rather than how fast it is, because you can adjust in DAW if you need to take out the latency in recording.

[TS-12]

Originally Posted by golden beers

having said this, i use USB. my stuff is tight enough. it's really about how uniform your midi timing is rather than how fast it is, because you can adjust in DAW if you need to take out the latency in recording.

Latency and jitter are two different things. Latency can be adjusted by shifting tracks earlier a few milliseconds. Jitter can't be adjusted by shifting notes earlier or later, because notes are sometimes on time and sometimes late

[Jauqq]

Originally Posted by TS-12

Latency and jitter are two different things.

Often see the two confused..
Latency is a delay from when you want something to happen until it actually happens, and jitter is small, random variations in timing. Latency can be compensated for, but jitter can not be.

[controlvoltage]

hey where's VelvetAcidChrist when you need him, he could really clear up any confusion with the terminology in this thread.

Thanks Altitude for posting thorough tech specs and the recommendation; nice one

[Yoozer]

It's just a crying shame that one has to solder this themselves, apparently. You could knock MOTU out of their monopoly with these things.