VOGONS


HardMPU, anyone?

Topic actions

Reply 280 of 608, by ab0tj

User metadata
Rank Member
Rank
Member

The parts to make kits and assembled units should be here this weekend. I'm going to get one soldered together and test before I start taking orders, so keep an eye out here early next week if you're interested.

DigitalBrains wrote:

Could you please post your Mouser saved cart, by way of detailed BoM? I can get parts cheaper than usual, with the downside that it takes a lot more time before I get them. So I'd like to order them.

Ok, the project is saved here: http://www.mouser.com/ProjectManager/ProjectD … ssID=b6acb4ad1b
Note that the saved project contains everything needed except a MIDI cable to chop up and solder a cable together with, and obviously the PCB.

Reply 281 of 608, by smbaker

User metadata
Rank Member
Rank
Member

Those look really nice. I'm surprised at the quick turnaround.

I ended up building my own (what's wrong with me? I can't just wait a week or two???) based on the schematic checked into the github repo, and it worked. Not only that, but it worked perfectly on the first try. Very nice job with the schematic, it was easy to follow. I programmed the atmega with minipro, same programmer that stamasd used (all the discussion of fuse issues had me worried, but I just ensured the fuse bytes were the same value as ab0tj's screen shot).

Reply 282 of 608, by ab0tj

User metadata
Rank Member
Rank
Member

I put together a unit this weekend and tested, all appears to be fine. That means I can start shipping them off to you guys. PM me to put in an order. Prices are $15 USD for a PCB, $60 for a kit (everything you need is included, even the MIDI cable, but you solder it together) or $100 for a fully built and tested HardMPU with cable. Shipping to the US is $7.20, PM me to discuss international shipping. Please remember I am not a business so I may not be able to ship your stuff immediately.

Warranty (because I know someone will ask)... I'd have to jack up pricing considerably to offer a "money back guarantee", but I will offer any remote help I can or I will diagnose and fix your unit provided you:
- Have completely built the unit if it is a kit or bare board.
- Pay for shipping both ways and any necessary repair parts.

P.S., I tried to find forum rules, and basically just came up with "don't be a jerk"... So if selling these here is not OK, please let me know. 😁

Reply 283 of 608, by ab0tj

User metadata
Rank Member
Rank
Member
smbaker wrote:

I ended up building my own (what's wrong with me? I can't just wait a week or two???) based on the schematic checked into the github repo, and it worked. Not only that, but it worked perfectly on the first try. Very nice job with the schematic, it was easy to follow. I programmed the atmega with minipro, same programmer that stamasd used (all the discussion of fuse issues had me worried, but I just ensured the fuse bytes were the same value as ab0tj's screen shot).

Just saw the YouTube video. Very cool, and I'm glad it worked out for you "out of the box". Now I have a reason to dust off my Xi8088 and play some KQ4 😁

Reply 284 of 608, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

I just watched it as well. Is your Yumcha USB MIDI adapter reliable or does it have an inadequate buffer for full SYSEX commands?

All hail the Great Capacitor Brand Finder

Reply 285 of 608, by DigitalBrains

User metadata
Rank Newbie
Rank
Newbie

I'm going over the Mouser saved cart (thanks for that!) and I happened to notice something. The 18 pF capacitors you use to load the crystal have a lead spacing of a mere 2.5 mm (about 0.1"). When I look at the photo of the PCB, the lead spacing looks to be 5 mm or 0.2" (both lead spacings should fit). So perhaps you want this capacitor instead:
810-FG28C0G1H180JNT6
or if you're rather attached to the colour 😀:
594-K180J15C0GF5TH5
or if you must have 0.2" lead spacing:
80-C317C180J5G
... or pick one yourself, they've got enough choice 😁

I did this using components that were stocked for delivery to The Netherlands, by the way. Please check for your area.

<edit>
Just to clarify to people who might be worried they ordered the wrong components: Don't worry, you can fit those 2.5 mm capacitors on the PCB. It just looks slightly less neat, that's all.
</edit>

<edit 2>
Changed the one with the 0.2" pitch... old one was 200 V with thicker leads, bit overkill for loading a crystal.
</edit 2>

Reply 286 of 608, by ab0tj

User metadata
Rank Member
Rank
Member

Yep, I had been meaning to fix that. Thanks for the list of alternatives. The ones in the cart work just fine, but it would be convenient to have the "right" spacing. I have updated the saved cart.

Reply 287 of 608, by smbaker

User metadata
Rank Member
Rank
Member
ab0tj wrote:

Just saw the YouTube video. Very cool, and I'm glad it worked out for you "out of the box". Now I have a reason to dust off my Xi8088 and play some KQ4 😁

It definitely completes the experience to run a homebuilt HardMPU in a homebuilt PC. Did you experience any stability issues with your Xi 8088? Myself and a few other guys here on the forum have been working on Xi 8088 stability issues for a couple months.

gdjacobs wrote:

I just watched it as well. Is your Yumcha USB MIDI adapter reliable or does it have an inadequate buffer for full SYSEX commands?

I haven't evaluated it in detail. I did hear sound effects with SQ3, and that seems to imply it was able to some SYSEX commands to set up the custom sounds. Is there a particular problematic test I can run that would show issues? I just grabbed whatever USB-MIDI adapter was cheapest on Amazon at the time, that wasn't flooded with low reviews.

Reply 288 of 608, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
smbaker wrote:
gdjacobs wrote:

I just watched it as well. Is your Yumcha USB MIDI adapter reliable or does it have an inadequate buffer for full SYSEX commands?

I haven't evaluated it in detail. I did hear sound effects with SQ3, and that seems to imply it was able to some SYSEX commands to set up the custom sounds. Is there a particular problematic test I can run that would show issues? I just grabbed whatever USB-MIDI adapter was cheapest on Amazon at the time, that wasn't flooded with low reviews.

Yeah, I don't have one, but apparently an adapter with the problem in question will corrupt any SYSEX message longer than four bytes. The driver can be patched to fix this for those with source access.
https://lists.debian.org/debian-kernel/2011/05/msg00119.html

Of course, the device itself is usually incomplete in an exercise of extreme cost cutting.
https://karusisemus.wordpress.com/2017/01/02/ … w-to-modify-it/

All hail the Great Capacitor Brand Finder

Reply 289 of 608, by smbaker

User metadata
Rank Member
Rank
Member
gdjacobs wrote:

Of course, the device itself is usually incomplete in an exercise of extreme cost cutting.
https://karusisemus.wordpress.com/2017/01/02/ … w-to-modify-it/

Wow! Well now I'm going to have to cut mine open and have a look. I haven't done any MIDI IN with it, so I wouldn't have noticed any issues from the lack of the optocoupler and diodes. It's silly they made such a cost-cutting measure, but not entirely unexpected.

Reply 291 of 608, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

I'll probably test the build with a DE15 jack and a gameport adapter as I hate soldering D-Sub plugs.

All hail the Great Capacitor Brand Finder

Reply 292 of 608, by DigitalBrains

User metadata
Rank Newbie
Rank
Newbie
gdjacobs wrote:

I'll probably test the build with a DE15 jack and a gameport adapter as I hate soldering D-Sub plugs.

The gameport MIDI interface is supposed to be electrically quite different, with just TTL-level signals rather than the usual MIDI interface (which I suppose is a current loop).

The ATmega works with TTL-level signals, so you could just hook the DA-15 directly to the microcontroller (you probably know, but note that DE-15 is a VGA connector, not a gameport connector, don't buy the wrong one 😀). Although it's arguably a good idea to use resistors in series so the ATmega survives shorts, so you could view R4 in that light. They're pretty sturdy, though, they don't short out easily.

ab0tj wrote:

To all who have already bought a kit, here are the build instructions: https://github.com/ab0tj/HardMPU/raw/master/B … e%20HardMPU.pdf

Wow, that's an extensive build description, good going! I reckoned the silk screen was already good enough!The only things that aren't fully defined on there are D1 and J1, and there's only one diode in the whole design 😁. Hmmm, although you could argue about which pin is the marked pin on a tantalum cap, I suppose.

Reply 293 of 608, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
DigitalBrains wrote:
gdjacobs wrote:

I'll probably test the build with a DE15 jack and a gameport adapter as I hate soldering D-Sub plugs.

The gameport MIDI interface is supposed to be electrically quite different, with just TTL-level signals rather than the usual MIDI interface (which I suppose is a current loop).

Well, the adapters do current loop conversion internally. The required location is indicated in the electrical schematic. I'd just bodge in ground and signal wires to the jack.

DigitalBrains wrote:

The ATmega works with TTL-level signals, so you could just hook the DA-15 directly to the microcontroller (you probably know, but note that DE-15 is a VGA connector, not a gameport connector, don't buy the wrong one 😀).

My bad. it's a cold day. My brain's not starting so well!

All hail the Great Capacitor Brand Finder

Reply 294 of 608, by DigitalBrains

User metadata
Rank Newbie
Rank
Newbie

After the supposed gameport-to-MIDI adapter circuits I've seen people propose on the internet, I thought it couldn't hurt to point out what hopefully was already fully known 😀.

Truly, there are some outright dangerous circuits out there; I wouldn't be surprised if some would cause the magic smoke to leave a chip on the soundcard. There was one that used MIDI-in as a signal ground 😁 (not that that would fry anything, it's just unbelievably stupid).

<edit> Might have been +5 V actually rather than ground, can't say for sure.</edit>

Reply 295 of 608, by smbaker

User metadata
Rank Member
Rank
Member
gdjacobs wrote:

I'll probably test the build with a DE15 jack and a gameport adapter as I hate soldering D-Sub plugs.

Personally, I bought some crimp DB9s just for this project because I too hate soldering D-Sub plugs. Also has the advantage that if one messes it up, there's a $3 tool that lets you non-destructively remove the crimp pins and do it right.

Reply 296 of 608, by ab0tj

User metadata
Rank Member
Rank
Member

Personally I hate the crimp DE9 connectors and have no problem soldering a DE9 when using a set of "helping hands", but that's just me. Maybe next time I order kit parts I'll get both styles and offer the option of either one.

Reply 297 of 608, by ab0tj

User metadata
Rank Member
Rank
Member
smbaker wrote:

It definitely completes the experience to run a homebuilt HardMPU in a homebuilt PC. Did you experience any stability issues with your Xi 8088? Myself and a few other guys here on the forum have been working on Xi 8088 stability issues for a couple months

I haven't experienced any issues, though I can't say I've logged a lot of hours with my Xi8088 either.

Reply 298 of 608, by DigitalBrains

User metadata
Rank Newbie
Rank
Newbie

And another HardMPU is now in existence! It's working fine, as expected. Note that not all parts you can see in the picture are from Mouser.

20170923_190455.jpg
Filename
20170923_190455.jpg
File size
331.29 KiB
Views
1022 views
File license
Fair use/fair dealing exception

Regarding those parts from Mouser, here's some more comments:

  • I count 11 100 nF capacitors, but the Mouser cart has 12.
  • The 33 µF capacitor in the Mouser cart has a lead spacing of 5 mm, but the PCB has a 0.1" lead spacing. Looking at the capacitor, probably the only difference is the way the leads are bent; I just straightened them. A Mouser alternative part would be, e.g., 581-TAP336K010SCS.

Furthermore, some parts were out of stock. Instead of the picoPower ATmega1284P, I got the non-picoPower ATmega1284. The only difference is that the ATmega1284P uses less power. Completely irrelevant for this application.

I couldn't find anything regarding how to program the fuses in the ATmega, so I simply took sane settings:

avrdude -p atmega1284 -U efuse:w:0xff:m -U hfuse:w:0x99:m -U lfuse:w:0xe7:m

This instructs avrdude to set the fuses to:

  • Disabled brown-out detector (BOD)
  • Full-swing crystal oscillator, fast rising power

Other fuses at default or obvious setting. If I know my ISA, /RESET is kept asserted until power is stable. This should mean that you could even completely remove the startup delay in the AVR, but since this is "not recommended" according to the datasheet, I took the conservative choice. It doesn't matter anyway, I just like doing things thoroughly, so it did cross my mind.

Finally, here's a pic with the card inserted into my Tandon PAC Plus 286:

20170924_132306.jpg
Filename
20170924_132306.jpg
File size
466.43 KiB
Views
1022 views
File license
Fair use/fair dealing exception

It joined the Adlib, Gemini VC-001 EGA card and Novell NE2000 already there (real NE2000, not compatible). The Datapac controller is not currently inserted because none of my Datapacs start up correctly. All the drivers and utilities for Datapac management are of course on one of the Datapacs 🙁.

I'm really happy with my HardMPU!