ChromaTalk Archives: May 2015
- Re: 2x40 characters HD44780 (4 messages)
- Channel Mobo Woe (4)
- Re: Tapper/'phantom' button presses (was: CC+ arrived, looks great, except for....) (6)
- spare power supply
- Burning CC+ 218 Firmware (2)
- Sound Banks from Video (9)
- Chroma 21030454 For Sale By Owner
- multimode filter (formants) (5)
- OT: ARP 4075 VCF "replica"
- Help w/ Chroma Sysex Issue (3)
Re: 2x40 characters HD44780
Go to first message in thread, April 2015
min struct · Sun, 3 May 2015 01:47:17 +0200
Sorry if this question has already be answered. Could this I2C character oled module work ? i was not able to identify its controller Op202i :- white : eBay item #161575128497
- green : eBay item #131185021178
thanks
David Clarke wrote 2012/11 [see thread an additional display for the Chroma]:
for reference : parallax support "ASCII characters Dec 32-127"There's no 'protocol' per se. The data coming from the CC+ is essentially raw serial data (e.g., 1 Tx line running at 19200, sending ASCII characters).
If you were look at the data with a communications program (hyperterm, TeraTerm, tip, etc.) you'd essentially see exactly the same data you see as text on the display. The only slight additional information is that there are a couple control codes also sent to the display which tells the display to "go back to the top" ( move to line 0, pos 0 ), so that the data always starts at the top and doesn't just continuously scroll off the bottom.
Whether it is configured for the SparkFun controller or the Parallax display, the CC+ always outputs ASCII characters to the serial line.
- 27979-Parallax-Serial-LCDs-Product-Guide-v3.1.pdf
- 27976-Serial-LCD-BS1-BS2-Code-RevF.zip
- 27976-Serial-LCD-SXB-Code.zip
- 27976-Serial-LCD-Javelin-Stamp-Code.zip
- 27976-Serial-LCD-Propeller-Code.zip
min struct · Sun, 3 May 2015 02:05:06 +0200
Could this board helps the compatibility of the discussed MCOT42005AX-EYM display module with the CC+ ? MIDAS MCRS232B DEV BOARD, RS232 FOR OLEDS 19200 BPS 1663677.pdf
reference : MCOT42005AX-EWM OLED, 4X20, TAB, WHITE, MULTI I/F 1722507.pdf
Roel D. [21010092] · Mon, 4 May 2015 11:01:25 +0200
It is probably fairly easy to adapt the code. Even more so since the library for the display already exists.
At the moment I'm too busy to do it myself. I might have time in July, but you should remind me of this conversation... :o)
My dream would be a graphical oled interface using the computer interface port, but for this I will probably need a few months (or years the way my life is going now....)
Cheers!
Roel
min struct · Mon, 4 May 2015 15:24:42 +0200
Count on me for the reminder ;)
On the oled parallel display side, I found this arduino article but I'm not able to open the "read more" link with my smartphone Midas 2004 OLED Character Display
Channel Mobo Woe
Jeffrey D. McEachin [21030073+] · Sun, 03 May 2015 23:35:46 -0700
After a 2 1/2 year break, I'm once again attempting to resuscitate #21030425 so that I can lend it to a friend. It originally went down with PSU problems. I replaced that w/ a MAP80 per Don Tillman's instructions (thank you, Don!). I also had to repair a few channel boards. For good measure, I installed a CC+, updated its firmware to V218, and hooked up an LCD temporarily to help diagnose channel board problems.
That leaves me with one big problem, which I believe is on the CMB: channel boards 3 & 7 won't tune. I've tried 4 or 5 known-good boards in each slot, but they always fail. A Set Split 40 returns CeE2 CfE2 C6E2 C7E2 - oscillator scale factor computation error on both channels of boards 7 & 3.
To rule out the I/O board, I hooked up the CMB to my other Chroma (glad the ribbon cable is pretty long). Same results.
I haven't seen anything odd in some cursory probes of the CMB with an oscilloscope. Any ideas as to what I should be looking for?
Thanks!
Jeffrey
David Clarke [21030085++] · Mon, 4 May 2015 07:21:58 -0300
.... channel boards 3 & 7 won't tune. I've tried 4 or 5 known-good boards in each slot, but they always fail. A Set Split 40 returns CeE2 CfE2 C6E2 C7E2 - oscillator scale factor computation error on both channels of boards 7 & 3...
By default, the tuning code will run until the first major error, then it will give up and deem the voice board as problematic.
As you're only reporting one error per channel, I'm assuming that the [Set Split][40] result noted above is just with a standard boot.
Alternatively you can use [Set Split][31] to tune the Chroma. By doing that, all elements of the channel cards are checked and all failures will be logged.
You can then issue [Set Split][40] again, and you should see a longer set of errors. This will let you know more about what was/wasn't seen by the system (e.g., if it is only problems with the oscillator scaling, or offset and filter-related items too).
I haven't seen anything odd in some cursory probes of the CMB with an oscilloscope. Any ideas as to what I should be looking for?
Given that it is channel board 3 and 7 that seem affected - and these only differ by one binary bit - my first thought is that one of the decoder lines that should select data for board 3 or 7 is ending up selecting them both at the same time. A quick check here would be whether the behaviour of one slot is/isn't impacted by the presence/absence of a card in the other slot (e.g., does slot 3 only fail, or fail differently with [Set Split][31], [Set Split][40], if the slot 7 card is removed).
Jeffrey D. McEachin [21030073+] · Tue, 05 May 2015 00:54:50 -0700
At 03:21 AM 5/4/2015, David Clarke wrote:
As you're only reporting one error per channel, I'm assuming that the [Set Split][40] result noted above is just with a standard boot.
Nope, I did a Set Split 31 first. The user manual is incorrect on this point - it stops at the first major error per channel BOARD, not the first error per channel. After boot it displays C6E2, CeE2. After Set Split 31 it displays CeE2 CfE2 C6E2 C7E2 (not sure why the order is reversed). Also, it sporadically shows an E4 on one of the channels, but I don't think that's significant.
Given that it is channel board 3 and 7 that seem affected - and these only differ by one binary bit - my first thought is that one of the decoder lines that should select data for board 3 _or_ 7 is ending up selecting them both at the same time. A quick check here would be whether the behaviour of one slot is/isn't impacted by the presence/absence of a card in the other slot (e.g., does slot 3 only fail, or fail differently with [Set Split][31], [Set Split][40], if the slot 7 card is removed).
I had the same thought and already tried that. Boards that are plugged in fail the same way regardless of whether the other board is there or not.
I also verified that all channels will play patches in 16 channel mode, though the bad ones don't track well (as should be expected), and channel f makes random popping sounds. A clue, I'm sure.
David Clarke [21030085++] · Tue, 5 May 2015 09:51:51 -0300
... The user manual is incorrect on this point - it stops at the first major error per channel BOARD, not the first error per channel.
The manual currently says "...you will only receive information on the first error caught on a channel." Insofar as it goes this is correct - but I agree that it would better to say "...caught on a dual channel board" as during standard tuning it will fail on the basis of a 'board'.
... I did a Set Split 31 first.
This puts a different complexion on things as it implies that the other test points passed. That means that the equivalent tests for the filter scaling and computation were OK.
Also, the fact that the oscillator high and low end measurements didn't fail outright says that the oscillator was nominally able to be controlled (which you've also confirmed via the practical tests).
As it was just the 'scale factor computation' which failed, it implies that what the system perceived as the 'measured' oscillator values were just not close enough to expected range for the math to otherwise figure out a valid scale factor. That could mean the cycle started too soon, too late, something gliched to trigger the detector sooner, etc.
For an oscillator measurement, the voice card has 'STB0' or 'STB1' data (for voice A or B, respectively) to select:
Wave Shape = Pulse Filter Resonance = 0 Filter = Low Pass
The oscillators are enabled, and the filter is opened wide.
The measurements are then done.
When it comes time for the filter, the on-card oscillators are disabled (via the OSC_EN line) - and then the work is done on the filter.
So - bad things could happen if items like STB0/STB1/STB2 were delayed/wrong. Ditto for SHA.
If you have a digital storage scope then you could likely capture the attempts to tune the 6 octaves apart for that channel card and see what you see (e.g., do you see a premature glitch on one, making the period measurement incomplete, is there an odd waveshape, etc.). You could also compare the relative strobes coming into the card (to others slots that are OK).
That said - as much as I generally don't recommend it - the 'fastest' approach may simply be through selective component replacement.
Earlier testing showed that the problem follows the channel motherboard so the implication is that there's something on there that's weak/failing. That means you're going to end up changing something on there anyway, meaning you're going to have to remove the board regardless.
As the parts in question are all standard, inexpensive items - and it board will already be out - you might just want to replace the most likely items at once.
You can rule out items that are common to all boards (e.g., Z3, Z6, all the output amps, noise generator, items only servicing other boards - e.g., Z13, Z11, etc.)
That would then leave you with a short list of components to consider (e.g., Z5, Z7, Z9, etc.)
Total brute force wuold be to look at the chips associated with the data for a whole channel. For instance, for Channel 7 that might be:
Z1, Z4, Z5, Z7, Z8, Z9, Z10
After boot it displays C6E2, CeE2. After Set Split 31 it displays CeE2 CfE2 C6E2 C7E2 (not sure why the order is reversed).
[Set Split][50] is expected to order them in reverse - but [Set Split][31] should order them from lowest to highest.
Re: Tapper/'phantom' button presses (was: CC+ arrived, looks great, except for....)
Go to first message in thread, February 2015
Frank McGing [21010198] · Mon, 11 May 2015 01:14:02 +0100
On 05/02/2015 04:36, Eric Frampton wrote:
I’m running into a similar behavior on #0501. Ever since I re-assembled it, when it’s cold it’ll start acting as if various buttons on the bottom row of the left-hand membrane panel are being pressed (most often Link Unison). The tapper will fire (sometimes rapid-fire), and the corresponding LED will blink. The odd thing is, if I physically disconnect that membrane panel, the unit will act the same way. Odder (?) still, when I raise the front panel to troubleshoot the I/O board, the problem always goes away as long as the panel is raised - which makes troubleshooting challenging.
Assuming the membranes are all connected, and with the front panel closed, if I tap on it it’ll act like random left-hand membrane buttons are being pushed. Then if I knock hard enough, it’ll usually freeze the whole machine up, requiring a power-cycle to reboot.
PSU voltages are spot-on.
My machine never ran with the original CPU board in it while I’ve had it, so I can’t speak for pre/post CC+ behavior.
Wonder if this is one or more microscopic trace cracks, or if there’s something on the edge of getting grounded out in the way the I/O board is mounted?
Hi Eric
Did you find a solution for this issue, or does anyone have any ideas?
My Chroma has been sleeping for a few months, and I woke it up for a session yesterday to find that it's doing exactly what you described. Link Unison is also the switch that seems to be triggering the tapper, along with (sometimes) Edit A.
Frank
Paul DeRocco [21030230] · Mon, 11 May 2015 08:23:17 -0700
Did either of you measure the supply voltages at the I/O board, or just at the supply? It sounds like a bad connection somewhere. I'd unplug and replug everything in sight. It might also be useful to look at the supplies with a scope, to see if there's any ripple on them. Broken traces or shorts are certainly possible, but less likely.
Eric Frampton [21030501] · Mon, 11 May 2015 11:59:55 -0400
Hi Frank -
I haven’t worked on this since re-assembling the instrument last Winter, and I’m on the road with the band Boston all summer (come see us!), so it’s going to be a minute before I can work on this again. The minute I discover something definite I’ll let you know.
e
Frank McGing [21010198] · Thu, 14 May 2015 00:00:50 +0100
Thanks Paul! I haven't had time to work on it since, but will definitely do as you suggest. Hopefully it is just a bad connection.
Frank McGing [21010198] · Fri, 29 May 2015 22:00:27 +0100
I've attempted to investigate this issue again, but it seems to be worse that I thought. I unplugged and reseated all connectors on the I/O board but there was no improvement. I also used a pencil eraser on the membrane tails, but again, no improvement.
Supply voltages at the I/O board are 5.05V analogue and 4.97V digital. Unfortunately I don't have a reliable scope, so I measured with a DVM.
- Pressing any of the Link switches makes the tapper go crazy and causes erratic flickering of the LEDs for the four Link switches, and Store also flashes sometimes. Link Unison keeps 'attempting to switch itself on'.
- Pressing Param Select seems to kill the Chroma for around 4 seconds! It goes dead and comes back to life with all LEDs on for a moment, then the panel LEDs return to the settings that they were on, except with Param Select active. If I press it again, the same thing happens.
- Edit A doesn't seem to be working at all - nothing happens when I press it.
I have a session tomorrow, and I've just discovered that if I unplug J4 on the I/O board, I can at least flick through the presets and play the Chroma without it going completely crazy. So that's what I'll do for now.
Any suggestions would be very much appreciated.
Frank
Paul DeRocco [21030230] · Fri, 29 May 2015 18:26:31 -0700
There's no J4. Did you mean J14, the leftmost membrane switch connector? If so, that suggests a problem in the switch scanning circuit, or possibly the left membrane switch panel itself. Since the STB0-6 lines on pins 1-7 of J14 also go to the right panel on J15, then they're probably okay on the board (that's assuming that all 50 of the numbered switches work when you unplug J14). That leaves the SW5-7 lines on pins 8-10, or the panel itself.
You can try things like leaving J14 unplugged and touching a wire between one of pins 1-7 and one of pins 8-10, to see if that invokes a clean switch press, or results in the same haywire behavior. If the former, then there's a problem with the membrane switch; if the latter, then the problem's on the board, and could be Z16 or Z17.
If it turns out to be the panel, it could be some abrasion through the insulation on the ink side of the tail, allowing it to make contact with the metal panel. I would think that would be pretty easy to fix by sliding a sliver of plastic in between the tail and the metal.
Go to next message in thread, June 2015
spare power supply
Steve Lodder [21010043] · Thu, 21 May 2015 15:22:42 +0100
here:
I recently replaced the power supply so have the one that was taken out surplus; anyone interested? it was working fine.... [email address removed]
Burning CC+ 218 Firmware
Ian Hamilton [21010150] · Fri, 22 May 2015 14:34:01 +1000
Hi,
I’ve had my Chromas with a tech for doe time, just got one back. He tried to burn the 218 firmware before he dropped it back but his attempts failed. Is there anything specific he should try?
Cheers, Ian.
David Clarke [21030085++] · Mon, 25 May 2015 16:13:05 -0300
There's nothing 'special' which needs to be considered so the focus would just be on looking to see what errors (if any) the burner is reporting, as well as ensuring that the contents of the 'burned' chip do verify successfully against the binary image.
Sound Banks from Video
Eric [21010243] · Sat, 23 May 2015 11:41:27 +0200
hello chroma users....maybe some of you could tell me where i can find those sound banks from the video...any help would be great thank you...here is the video
all the best eric
Buddy Casino [21010137] · Sat, 23 May 2015 12:16:27 +0200
Hi Eric,
these are just the good old factory presets. You can find them on the Chroma page [Patch Downloads].
Eric [21010243] · Sat, 23 May 2015 20:28:24 +0200
thank you buddy ...still missin a plug cable to transfer those factory presets into the chroma....best eric
Buddy Casino [21010137] · Sat, 23 May 2015 21:33:17 +0200
Hi Eric,
Don't u have the really great working CC+ Midi kit?
Man, that's a must! Then you load and save sounds via Midi sysex.
Go and get it, it's such on improvement.
Buddy
Eric [21010243] · Sat, 23 May 2015 21:43:33 +0200
no i dont have the cc+ kit....actually i wanted to order it but then decided to keep it in original condition beside the voltage supply which was changed into the newer one...still must registrate my chroma serial number is 21010243 ...i got it from a guy called nenad in vienna austria he was a guitarr player...i paid 5000 euros....then i took it to my technician nikolaus riehm who is kind of a living legend here in germany...totally nerd this guy and very friendly
David Hobson [21030506] · Sat, 23 May 2015 20:02:37 -0400
Hello, Blankowv.
I have a spare Syntech midi interface for the Chroma. It is in like-new condition, and you would need a midi cable.
Let me know if you are interested and we can work out the details. Regards.
~David Hobson.
Eric [21010243] · Sun, 24 May 2015 12:19:52 +0200
hello david thanks for your attention but i already got the midi interface from jl cooper electronics.....best eric
David Clarke [21030085++] · Mon, 25 May 2015 16:19:18 -0300
Loading patches with MIDI SYSEX
... still missin a plug cable to transfer those factory presets into the chroma....
While the original Chroma cassette cable (e.g., Info on Chroma Cassette Cable, January 2001) had small connectors for portable cassette decks of the day, most times now we're not needing to worry about control of an external recorder motor and so that full cable is really not needed.
Similarly, if it is only desired to load images into the Chroma (vs. saving them), then it is just two signals which are interest. You could use the information at the link above and a MIDI connector to make a quick cable to allow access to the cassette interface.
Best regards,
David Clarke
Jesper Ödemark [21010135] · Tue, 26 May 2015 08:36:34 +0200
Re: Loading patches with MIDI SYSEX
I built my own, but with a breakoutbox with female connectors marked "in" and "out". I found this more useful than fiddling with male connectors. I might bring those out as sockets on the front when I finish my programmer...
Chroma 21030454 For Sale By Owner
Chris Ryan [21030691] · Sun, 24 May 2015 16:56:53 -0700
Chroma 21030454 is for sale. John Simmons writes:
"I have been able to significantly improve it using [the site], so the Chroma for sale comes with the CC+ with midi box and SPSU upgrades installed, and the Polypressure kit boxed and ready to be installed. I’m also selling it with a BCR2000 and the latest overlay (which has not yet been attached) and a set of the Chroma pedals. It has been regularly serviced but has the following issues which would need correcting to bring it to full working order and good cosmetic condition all round: Voice board 4 is not working; The thin wooden strip in front of the keys needs replacing (but this does not affect the working order of the Chroma at all – hence I have never replaced it). The pedals are not working.
“The only person who has ever worked on the Chroma is James Walker of Synth Repair in Stoke, (so I’ve never allowed anyone to dig around inside who isn’t basically the best in the business.)
"I can arrange for courier but would obviously recommend collection in person from Tunbridge Wells (or I may be able to deliver within the UK).”
John is looking for £3500 for his Chroma. A couple of pictures:
He also has some other items available, and is open to offers for a quick sale: Roland SH3a - £375; Roland MC202 - £475; Roland TR606 - £275; Roland 106 (not working – voice chip requiring ‘wash’) - £325; Roland Juno 60 (not working – a tech minded friend says this is a PCB fault) - £525; EML home built semi-modular oddity - £995; Oberheim OB8 (excellent cosmetic condition but not working!) - £1600; Korg Wavestation - £275; Novation Basstation - £125; Yamaha DX21 - £125; Arp Axxe and Little brother modified combo in case with legs (with Saint Eric being serviced/repaired and ready soon) - £995; 2 ADR compex Compressor/Gate/Expanders lifted form my old Neve console (they would need repair and power supplies built/racked) - £995; Lexicon LXP1, LXP5, PCM 70, MRC remote control, PCM80 (needs new display which proved hard to find at a reasonable price) – Offers each or £795 for all; Midas Venice f24 console used twice and covered ever since! - £995; Loads of odd fx hardware like Drawmer 241 and the usual old bits like quadraverbs etc.. – Offers?; Matching NS10’s with 2 x Yamaha PM2500 amps - £575; Korg Lambda (literally totally knackered cosmetically!!!) – Offers?”
multimode filter (formants)
Frank Bonarrigo · Wed, 27 May 2015 09:46:40 -0400
Hi, this is the only synth forum im subbed to and my question could still be relevant. I'm reading Allen Strange's book [Electronic Music: Systems, Techniques, and Controls].
Has any one seen a multimode filter for isolating, boosting, carving formants for a synth? Do they exist? If so point me to one. please.
The book says you can isolate partials and imitate woodwind bodies. This would be cool to do with a Chroma. Or maybe the Chroma can already do this?
thanks
Heinz Weierhorst [21010276] · Wed, 27 May 2015 16:02:36 +0200
Hi Frank.
The Polymoog has three of them called resonators.
Heinz
Jeffrey D. McEachin [21030073+] · Wed, 27 May 2015 14:01:14 -0700
At 06:46 AM 5/27/2015, Frank Bonarrigo wrote:
Has any one seen a multimode filter for isolating, boosting, carving formants for a synth? Do they exist? If so point me to one. please.
The EQ on your Chroma's front panel is a very crude formant filter. Any multiband EQ will work, but for serious formant shaping, you'll want tunable bandpass filters with adjustable Q, and more of them.
Try an image search of formant filter, and you'll see numerous hardware and software implementations. Software formant filters are typically cheaper, quieter, and more powerful than hardware ones. If you'd prefer a hardware filter, there are many modules and pedals to choose from. I'd recommend playing with a free VST first to figure out how many bands are useful to you and what features you need.
At 07:02 AM 5/27/2015, Heinz Weierhorst wrote:
The Polymoog has three of them called resonators.
The Korg PS-3X00 synths also have 3 filter resonators. The Arp Pro-Soloist/Pro-DGX has several banks of resonators with a total of 14 fixed filters. I'm sure there a other synths that used them as well. I wish more synths had them.
Jeffrey
John Leimseider [21030434++] · Wed, 27 May 2015 21:18:38 +0000
The Synton Syrinx also has some great filters that can be used as formants, based on the CEM3350 filters [PDF], the same as on the Chroma.
Jesper Ödemark [21010135] · Thu, 28 May 2015 13:45:26 +0200
MAM RS-03 is my favourite tool for such operations. Three parallell bandpass filters in one 19" rack unit...
In addition to this you can test a filterbank. The MAM (again?!) AFB-8 is often very cheap out there and does a great job. There are others, by Analogue Solutions etc.
For best analogue woodwinds on the planet there is only one choice: <3 Realton Variophon Gig <3
OT: ARP 4075 VCF "replica"
Luca Sasdelli [21010226] · Wed, 27 May 2015 16:04:09 +0200
Hi all,
I'm about to begin a small production of the modules on subject. The reason is that they're a bit difficult to service and very hard to find. The 4075 was the definitive VCF for ARP Odyssey 2810-2813, Axxe 2310-23233, Omni, Omni-2, Pro DGX, Avatar and Quadra.
The filters I'm intended to build have the following features:
All components are THE SAME as the original one - no equivalents nor replacements used.
Matched transistors are manually selected for delta hFE < 10
All PCB track paths follow the original layout
All resistors are metal film with 1% tolerance (or better)
All filter cells capacitors are WIMA high-quality polypropylene with 2.5% tolerance
Top layer grounded silkscreen to improve noise rejection
Z1 mounted on high-quality socket for easy servicing
Timothy Smith's mods are applied:
- filter 12kHz limit extended up to 28kHz (it was a design miscalculation)
- a zener diode to stabilize local 6V power supply against fluctuations (latest mod)
- output capacitor is ELKO audio-grade electrolytic instead of tantalum one
Anybody interested?
Cheers
Luca
Go to next message in thread, June 2015
Help w/ Chroma Sysex Issue
Derk Reneman · Fri, 29 May 2015 13:57:38 +0200
Hi there,
I’m sending sysex patches to my chroma with Sysex Librarian (Mac OSX) - not very complicated as you might understand. So far all seems to work as should – the Midi Activity Indicators are blinking during transfer. And, new patches actually do become available. At the moment I’m sending a patch file to bank 4. Memory protection is turned off.
My CC+ unit is running firmware 216 I’m using a Presonus Firestudio interface with Mac OSX 10.7
However, only 12 patches are being received/stored. Strange, as I presume most factory preset files would contain 50 patches …
I’ve tried numerous patch files, different transfer speeds and settings, sending multiple messages, and read about all manuals i could find. It seems either i’m not understanding the correct way to go about sending patch files from my computer to the chroma, something is wrong with the machine, or there’s something wrong with the software.
Is there a friendly someone who can help me out?
Much obliged, Derk
Brian McCully [21030361] · Fri, 29 May 2015 07:31:59 -0700
Hi Derk,
If you dump 50 patches out of your Chroma, do they dump back in? Otherwise maybe try another MIDI interface. Sounds like you're covering the bases otherwise.
David Clarke [21030085++] · Sat, 30 May 2015 09:16:40 -0300
If you dump 50 patches out of your Chroma, do they dump back in? Otherwise maybe try another MIDI interface. Sounds like you're covering the bases otherwise.
Following up to Brian's note, you can attempt to dump 50 patches from the Chroma to the librarian by issuing [Set Split][36], [P23].
1) Assuming you can successfully receive a full bank into the librarian - I'd recommend going to one of the other banks on the Chroma (say, Bank 1) - and then sending that bank to the Librarian.
On the Chroma, switch back to Bank 4 - then look to send the bank you've just received back to the Chroma in Bank 4 and look to see what happens/doesn't happen.
If Bank 4 is now identical to Bank 1, then that implies the sysex transfers (in general) are OK and the original issue may relate to the sysex files.
If Bank 4 is only partially written or doesn't sound like Bank 1 at all, then the nature of how it does/doesn't compare to Bank 1 may tell us something. In this case the recommendation would be to see if there's an alternate sysex sender on hand that could be attempted - whether that from inside a sequencer program, another standalone app - or perhaps a different computer/environment.
Notes:
- The Chroma actually has 51 patches active at any one time - the 50 in the bank, plus the 'current' patch. I mention this because even after you've sent a full new bank to the Chroma may you'd still hear the current patch, perhaps implying that the bank hasn't been updated - but as soon as you select one of the patch numbers, it will defined the current patch to be that from the new bank.
- To confirm what is happening, a quick test would be to take the same patch and to store it to all 50 patch locations (e.g., Store, P1, Store, P2, etc...) In that way, once you've tried to do the load via sysex you can quickly/easily see if there's been any changes in the bank or not - and if so, if it is on sequential patch numbers, etc..
- When evaluating whether the patch has changed or not - it is recommended to do that via the Chroma's keyboard (vs. 'playing' it via MIDI). This is just to simplify the setup.
- Again, to simply the setup, the MIDI OUT from the Chroma can actually be disconnected during the test (just to make it even that much more certain that there's no influence from MIDI. Sysex should only require MIDI in to the Chroma).