The CH375USB.sys driver is very picky about which "formats" it accepts. For example even if I just use DOS 6 itself (on another device) to format the partition on the USB, CH375USB.sys will then hang while booting if the pendrive is plugged in. The MBR & the geometry fields in the BPB are the same (DOS won't change it...), so it is some other field in the BPB that causes the hang. Formatting using Linux's mkdosfs actually works , though (forcing the same geometry). Comparing the stock CF images, I think manufacturer uses mkdosfs (or derivative) to format them.

(In unrelated topic I noticed USB-capable award BIOSes actually look at the BPB of the first partition to guess the disk geometry of USB pendrives, which in retrospect actually makes a lot of sense. I always assumed it just tried to guess the geometry by searching the one that would result in pre-existing partitions being cylinder aligned...).

Is there a volume control for the internal opl3? I've tried the headphone jack with two pairs of headphones I have and its so quiet it might as well be turned off.

dukeofurl wrote on 2025-03-03, 01:09:

Is there a volume control for the internal opl3? I've tried the headphone jack with two pairs of headphones I have and its so quiet it might as well be turned off.

The two potentiometers hidden behind the cover at the bottom (in blue and marked "volume" in the PCB) control L & R volume.

I've had my pocket386for a while now & had a good play and a look at the design...
I've been experiencing the same battery issues other have reported on here, a random message popping up saying that the battery is low (about 2400mV) when there is loads of charge in the battery & it keeps running for ages, but I cant see if others have had some conclusion to this? Is this something everyone has experienced? Or is it a fault only present on some units?
The battery voltage reading looking odd when the unit is on charge I think is part of the design; if the schematics/PCB files are correct, when the +12V supply is connected it switches off the 3.7V rail from the battery by shutting off the power MOSFET Q3 by bringing its gate high Via D8 & R23. The 8051 (U40) monitors the battery voltage at the point after Q3 (MCU_ADC1), so this is floating/undefined when the +12V supply is connected, it may well rise through leakage through the regulator – though it does seem odd that it reads over 4V at times – this would make the regulator operate briefly!

The low battery warning are as if Q3 or the battery protection circuits are momentarily disconnecting the battery (possibly bad connection?), but not long enough for it to shut down (perhaps hold up time of the 3.7V to +5V reg & its input cap?) or maybe software issue in the 8051??
On the FnF5 OSD I do get corrupted characters flashing up...
I wondered if others found it was down to a faulty battery and got a replacement from the supplier?
I've emailed my seller, but found it hard to video the precise moment when the message pops up...

I've also had issues with the parallel port on mine; it looks like the design is very similar to a PS2 parallel port design: Bi-directional made with LSI ICs (the BIOS settings seem to be largely ignored/irrelevant!),
However on mine the LSI chips aren't the 74LS on the schematic, they have been substituted for 74HC, presumably to save power. The open collector 74LS05 IC U30 has also been changed for 74HC04 & so the pull up resistors aren't fitted. This is used on the port control lines (Pin 1, 14, 16 & 17 on DB25). Unfortunately this cant sink enough current for any of the peripherals/parallel port devices I tried to connect to, it could only pull down to about 1.3V . The best fix for this would be something like:
(a) to change out U30 for something that can sink more current (& fit the pull up resistors if required). I think it should be OK to mix and match logic series here by putting in a 74LS05, with the pull up resistors the logic high level should be OK for the 74HC240 U2 it is driving. Or if you can find such a mythical beast a 74HC06 should be ideal...
Or
(b) Re design a breakout connector PCB with either the ICs above on or discreet transistor open collector with another stage to correct the inversion. The designer has put +5V on the connector, so there are lots of options.

I haven't tested either of the above!!
I wanted to get mine going without modifying anything & with whatever I had to hand, so I put a circuit on strip-board inside a DB25 connector. As there is no power supply here & I didn't want to steal power off the signal lines, options are limited, I went for a PNP emitter follower on each control line (see attached, repeat circuit on all 4 control lines). This is a bodge!, it can only get to within 650 to 700mV of ground, the standards call for more like 0.4 or 0.5V. It scrapes by as the input low level is 0.8V (though it is meant to be Schmitt trigger) I can confirm this bodge worked for me on an Epson stylus 740 printer, and external CDROM HP cd writer plus7200 & interlink to another PC.

Jules_nerd wrote on 2025-03-25, 20:30:

I've also had issues with the parallel port on mine; it looks like the design is very similar to a PS2 parallel port design: Bi- […]
Show full quote

I've also had issues with the parallel port on mine; it looks like the design is very similar to a PS2 parallel port design: Bi-directional made with LSI ICs (the BIOS settings seem to be largely ignored/irrelevant!),
However on mine the LSI chips aren't the 74LS on the schematic, they have been substituted for 74HC, presumably to save power. The open collector 74LS05 IC U30 has also been changed for 74HC04 & so the pull up resistors aren't fitted. This is used on the port control lines (Pin 1, 14, 16 & 17 on DB25). Unfortunately this cant sink enough current for any of the peripherals/parallel port devices I tried to connect to, it could only pull down to about 1.3V . The best fix for this would be something like:
(a) to change out U30 for something that can sink more current (& fit the pull up resistors if required). I think it should be OK to mix and match logic series here by putting in a 74LS05, with the pull up resistors the logic high level should be OK for the 74HC240 U2 it is driving. Or if you can find such a mythical beast a 74HC06 should be ideal...
Or
(b) Re design a breakout connector PCB with either the ICs above on or discreet transistor open collector with another stage to correct the inversion. The designer has put +5V on the connector, so there are lots of options.

I haven't tested either of the above!!
I wanted to get mine going without modifying anything & with whatever I had to hand, so I put a circuit on strip-board inside a DB25 connector. As there is no power supply here & I didn't want to steal power off the signal lines, options are limited, I went for a PNP emitter follower on each control line (see attached, repeat circuit on all 4 control lines). This is a bodge!, it can only get to within 650 to 700mV of ground, the standards call for more like 0.4 or 0.5V. It scrapes by as the input low level is 0.8V (though it is meant to be Schmitt trigger) I can confirm this bodge worked for me on an Epson stylus 740 printer, and external CDROM HP cd writer plus7200 & interlink to another PC.

The attachment Parallel CNTL bodge.gif is no longer available

Replace LSI with SSI!

Jules_nerd wrote on 2025-03-25, 20:26:

I've been experiencing the same battery issues other have reported on here, a random message popping up saying that the battery is low (about 2400mV) when there is loads of charge in the battery & it keeps running for ages, but I cant see if others have had some conclusion to this? Is this something everyone has experienced? Or is it a fault only present on some units?

I do see it all the time. While charging for the reasons you mention it appears almost constantly; while discharging it is not yet clear to me why, but it does appear, but rarely. It tends to correlate with SoC (appears more often with lower battery v) , I'd bet it is 8051 or sampling issue, after all it doesn't shut down as you say.

I never got parallel to work but I didn't try very hard; I got serial to work at least to an FTDI; the BIOS on mine is a bit buggy and will crash on even a simple echo > COM1, but raw access does work.

Jules_nerd wrote on 2025-03-25, 20:30:

I've also had issues with the parallel port on mine; it looks like the design is very similar to a PS2 parallel port design: Bi- […]
Show full quote

I've also had issues with the parallel port on mine; it looks like the design is very similar to a PS2 parallel port design: Bi-directional made with LSI ICs (the BIOS settings seem to be largely ignored/irrelevant!),
However on mine the LSI chips aren't the 74LS on the schematic, they have been substituted for 74HC, presumably to save power. The open collector 74LS05 IC U30 has also been changed for 74HC04 & so the pull up resistors aren't fitted. This is used on the port control lines (Pin 1, 14, 16 & 17 on DB25). Unfortunately this cant sink enough current for any of the peripherals/parallel port devices I tried to connect to, it could only pull down to about 1.3V . The best fix for this would be something like:
(a) to change out U30 for something that can sink more current (& fit the pull up resistors if required). I think it should be OK to mix and match logic series here by putting in a 74LS05, with the pull up resistors the logic high level should be OK for the 74HC240 U2 it is driving. Or if you can find such a mythical beast a 74HC06 should be ideal...
Or
(b) Re design a breakout connector PCB with either the ICs above on or discreet transistor open collector with another stage to correct the inversion. The designer has put +5V on the connector, so there are lots of options.

I haven't tested either of the above!!
I wanted to get mine going without modifying anything & with whatever I had to hand, so I put a circuit on strip-board inside a DB25 connector. As there is no power supply here & I didn't want to steal power off the signal lines, options are limited, I went for a PNP emitter follower on each control line (see attached, repeat circuit on all 4 control lines). This is a bodge!, it can only get to within 650 to 700mV of ground, the standards call for more like 0.4 or 0.5V. It scrapes by as the input low level is 0.8V (though it is meant to be Schmitt trigger) I can confirm this bodge worked for me on an Epson stylus 740 printer, and external CDROM HP cd writer plus7200 & interlink to another PC.

The attachment Parallel CNTL bodge.gif is no longer available

If a precise ohmage for stable current sinking is unknown, replacement with a tuning pot might be a better fix. That's easily revised without a lot of rework with just a plastic wand.

Im not too sure what your driving (no pun!)at with the ohmage for current sinking? I tried some pull down resistors as a quick and dirty fix, but the drive capabilities of the 74HC is so poor, by the time youve got resistor value low enough to get a decent logic low voltage, it really drops the high outputs too much for reliable operation on my printer..
By all means try a pot, values around 820 Ohms or so seemed OKish with my printer (but yeah its very dependant on the pull ups in your printer, I guess thats what your thinking about unknown ohmage?).
But the emitter followers really did seem to work better - across the various devices I tried..
Dont forget, Ive just showed a circuit for one of the control lines, you'd need 4 pots....

javispedro1 wrote on 2025-03-26, 13:19:

Jules_nerd wrote on 2025-03-25, 20:26:

I've been experiencing the same battery issues other have reported on here, a random message popping up saying that the battery is low (about 2400mV) when there is loads of charge in the battery & it keeps running for ages, but I cant see if others have had some conclusion to this? Is this something everyone has experienced? Or is it a fault only present on some units?

I do see it all the time. While charging for the reasons you mention it appears almost constantly; while discharging it is not yet clear to me why, but it does appear, but rarely. It tends to correlate with SoC (appears more often with lower battery v) , I'd bet it is 8051 or sampling issue, after all it doesn't shut down as you say.

I never got parallel to work but I didn't try very hard; I got serial to work at least to an FTDI; the BIOS on mine is a bit buggy and will crash on even a simple echo > COM1, but raw access does work.

Yeah, an 8051 issue would certainly explain a lot... and the random characters I get on the OSD on mine (again do others have this issue? pic attached) doesn't inspire me with confidence...
OTOH mine also randomly looses BIOS settings when powered down but with a fully charged battery, as others have reported, sometimes just re-sets the time, sometimes looses everything; floppy present etc... - so that makes me think that somehow the battery output is dropping for some reason?
I've not tried the serial port on mine yet, does anyone recognise the internal rectangular PCB footprint U3 for COM2 on my serial PCB? Its also tracked to J2, the legend for which is interesting, M0SI? MISO? SCLK? - sounds rather SPI bus??? ADC???

I suppose some may be lucky with the parallel port - the device you are connecting to could have higher value pull up resistors than usual on the control lines. Despite the low noise margins of putting 4 PNP emitter followers on the parallel port control lines, it has been working reliably for me; large amounts of data read from external (Shuttle EPAT based) CDROM without error. But if I didn't mind modifying the Pocket 386 & I wanted to fix it 'properly' Id try replacing U30 with 74HC06, or 74LS06 & fitting the pull up resistors (R102 through R107) the74ls05 on teh schematics still cant sink enough current to meet the standards... As I said I've not tested that though...

A word of warning, the Pocket386 can be partially powered by a printer or other device connected to the parallel port, mine has survived this without any apparent damage, but it does make me nervous!!
If you switch of the pocket386 before switching off a printer connected to it, it remains partially powered and hangs (screen goes off, but if you switch back on the screen looks the same as it was before you switched it off...). If you switch on a printer before the Pocket386 it doesn't boot, you get weird behaviour...
Presumably this is current from the printer going though I/O protection diodes (or maybe output FET) to the supply rail on the parallel port chips, which is why Id be careful about the order of switching things on/off...!!!

Sounds like a set of diodes on GND and +5v need to be added, so that attached devices cannot backfeed power.

javispedro1 wrote on 2025-03-26, 13:19:

I do see it all the time. While charging for the reasons you mention it appears almost constantly; while discharging it is not yet clear to me why, but it does appear, but rarely. It tends to correlate with SoC (appears more often with lower battery v) , I'd bet it is 8051 or sampling issue, after all it doesn't shut down as you say.

I never got parallel to work but I didn't try very hard; I got serial to work at least to an FTDI; the BIOS on mine is a bit buggy and will crash on even a simple echo > COM1, but raw access does work.

If you're getting the low charge message appearing while charging try a lower output charger. I switched mine out immediately for my high quality USB-C power brick and an appropriate USB-C to barrel jack. What I noticed is that my usual USB-C adapter which puts out as much amperage as it'd accept caused a constant low battery warning issue but a low power USB-C brick capable of a lower amperage didn't.

Also, I've meant to check what was going on there because that brick is capable of a (relative to what I assume that battery should charge at) *lot* of amps and I'm curious exactly how much power the Pocket 386 is drawing...

It turned out to be an intermittent solder joint on D8 – I only fixed it last night...
Its clearly had a knock, perhaps on assembly, maybe by a lever used to push out that side of the plastic case to get it over the line socket...
I guess this is more than likely only an issue on mine – but if I saw odd behaviour on another one when plugged in I'd check it, just in case there is something that someone is doing consistnatly on assembly...?
C22 solder joint doesn't look great either!
R22 is 510 Ohms rather than the 590 in the schematic.
Ive analysed this and 510 Ohms looks like a good compromise between ensuring the battery is switched off when 12V is applied & ensuring it switches back on when the 12V drops to a sensible value, given the full possible range of FET threshold and battery voltages.
With a lower input voltage than 12V its possible that the FET wont switch off the battery – exactly what voltage this happens at will depend on the threshold voltage of your particular FET and the battery voltage at the time...Perhaps this is what is happening with your USB brick?

I now get the OSD consistantly giving readings of 650 to 850mV when plugged in(Before when D8 was intermitent I was getting all sorts of readings as it switched the battery in and out of circuit...) & the 8051 code seems to recognise this as charging and displays battery % as 'CHG'
I don't get any more shutdown warnings pop up while it is plugged in, or at least I didn't see any after playing elite for a few hours!
I havent tried it just running on battery yet, but I dont expect to see an improvement in those circumstances...

Ive measured some of the power supply requirements, when the battery is fully charged it draws roughly 1.25A from the battery at DOS prompt (rising to over 2A when down to around 2.5V), it draws less(!) while booting up windows.. So it uses around 5W. The battery charge IC is set to charge at 1.5A (max 4.2V), so I guess you can work back to get a rough Idea what it would draw from your USB brick?

I also have this issue with the charger but it seems like it completely stops happening when the battery is fuller. When it happens it actually reboots the machine or crashes it every now and then so I suspect it's a real voltage drop. I will check those components mentioned above first chance I get.

Till then, I bought a PS/2 keyboard and a mouse. The mouse works as long as I disable the emulated mouse. But the keyboard stays with all three light lit up and no key does anything when pressed neither on the external keyboard nor on the laptop. If I remove it (with the computer still on) the laptop keyboard starts working again without even restarting the computer. Am I forgetting something? It's not a USB keyboard with a PS/2 adapter, unless it is internally like that.

In your case Id suspect the +12V to +5V regulator, or the MOSFET switch after it
It sounds as if maybe its struggling when loaded with both running the pocket386 & charging its battery at the same time (the battery charger runs from the +5V output of this regulator). I dont think its the +12V wall wart as the battery supply should kick in if that drops out (but you could monitor the 12V on pin46 of the ISA connector while its runnng to check that the wall wart output is OK?
ID check all the parts in the circuit around U22, including D4, D10, D20 & Q2

Though of course the vendor really should deal with these issues..... In my case I did get responses, but there was a little difficulty with translations...
And it was difficult for me to get videos of all the issues - in the end it was quicker for me to get out the soldering iron....
but the vendor was prepared to listen...

@Jules_nerd Hats off to you and your soldering! I couldn't have done that in a million years. Glad you could sort it yourself.

Thanks, though TBH the soldering is no great shakes for me, I'm a semi retired electronics engineer.... But lets not speak too soon, I haven't fixed all the issues yet!!
I still get the low battery messages popping up randomly when running of battery (the D8 issue was when running off mains adaptor). If this turns out to be an 8051 code issue then its beyond my ken to fix - I'm a hardware guy & mainly analogue at that. But you never know this old dog might learn some new tricks ?
I could be interested in doing some hardware projects for this, an internal covox PCB is in the pipeline (anyone interested? It will need soldering on to the parallel port connector & mod wires to the audio circuits to playback through the internal speakers) though I doubt Ill be happy with stopping there - I'd like some 'proper' soundblaster compatible digitised audio playback, so I might do an internal card, maybe ESS based for simplicity if i can get hold of a chip or maybe based on a soundblaster DSP chip, in which case Id make it cut down to just playback (im happy with mono 8 bit.. for games) and maybe a single joystick interface (for Xwing!)...
It'd need an extra connector with flying leads to solder on to the DMA lines as well as to the audio ccts...
Or possibly an external add on adaptor that fits on the back (possibly a new 3D printed case for the whole pocket 386 incorporating the add on card...), with 'real' PS2 connectors for mouse, keyboard, monitor etc....But I do rather like the compact form its currently in.
I guess one approach for either the internal or external adapter would maybe be a version of picogus - I guess basically an adapter card for a pi pico but I've not looked at that in detail - even if its allowed under license?

Ah OK, I should do a little more research before writing long posts!!... A Femto is more or less what I described as one of the external options...
And its a fiar bit more than an adaptor for a Pi....

Jules_nerd wrote on 2025-04-05, 15:03:

I now get the OSD consistantly giving readings of 650 to 850mV when plugged in(Before when D8 was intermitent I was getting all sorts of readings as it switched the battery in and out of circuit...) & the 8051 code seems to recognise this as charging and displays battery % as 'CHG'

Mine has always been like that forever (OSD showing CHG and a value in the 0600-0800 range) and yet the battery warning prompts still occur.
I just cannot imagine how it is really supposed to work with the voltage probe of the µc being where it is . I do not think it is avoidable.
At least the (blue/red) LED seems controlled by the charger IC and is reliable in my experience.