salfter/fluxengine
1
0
Fork 0
mirror of https://github.com/davidgiven/fluxengine.git synced 2025-10-31 11:17:01 -07:00
Code Issues Packages Projects Releases Wiki Activity
2,937 Commits 56 Branches 83 Tags
master
Commit Graph

69 Commits

Author SHA1 Message Date
David Given
6105e16b28 Don't invoke the vbs patch script (as it's gone). 2022-08-04 00:07:41 +02:00
David Given
22e65227fb Increase the generated pulse width to try and generate a stronger signal. This 
does seem to make Amiga disks read more reliably.
 2022-03-15 23:56:49 +00:00
David Given
d157b7b05d Make libusbp work on Windows; rework the Microsoft-specific bits of the 
firmware to match. This does at least allow us to get rid of the patcher.
 2021-12-12 17:25:24 +00:00
David Given
5f4f2f10d7 Update FluxEngine components and rebuild firmware. 2021-05-25 19:11:32 +01:00
David Given
c78ed2c6ad Add the TK43 pin, which goes low when seeking to track 43 or above. 2020-09-10 21:48:30 +02:00
David Given
972c8c6b61 Fix off-by-one sampler error, so now the clock rates are right again. 2020-04-03 22:27:33 +02:00
David Given
64694580cd Remember to bump the protocol number after the bytecode change. 2020-04-03 21:46:51 +02:00
David Given
1b48ea20c4 Remove the cruncher. 2020-03-20 00:06:07 +01:00
David Given
1a6c6b5420 The bandwidth tester now tests bandwidth in both directions. It looks like my 
default USB port only gets about 500kB/s write bandwidth. However, when plugged
into a port with 850kB/s, I still get underrun errors...
 2020-02-17 23:58:40 +01:00
David Given
ef4eff0195 So writing now works, but only if USB DMA is enabled. But that breaks reading. 
I've worked round this in a simple but brute force manner and it now looks as
if reading *and* writing work, more or less. There does still seem to be the
odd bad sector when writing 1440kB disks.
 2020-02-17 21:41:01 +01:00
David Given
df8d45bf66 Rework the output fifo to be a bit more correct about the sync signals, which 
in turn allows the sequencer to lose less time --- this gets the effective
clock rate down to about 1.01us. However we still seem to lose the last sector
on 18-sector disks and there are some disk reads so something is still wrong.
 2020-02-17 00:13:13 +01:00
David Given
933ffe7ab4 Find and attempt to fix a memory corruption error when sampling --- if the next 
fragment arrives from the sampler before usbbuffer has finished being
transmitted via USB, it'll get overwritten. I've disabled DMA USB to make the
code easier to understand and made sure that we flush things more rigorously.
This may help the weird pipe errors, too.
 2020-01-27 21:40:27 +01:00
David Given
d77841c3b7 Add the ability to fake the index pulse source, allowing old drives to be used 
with FluxEngine.
 2020-01-12 01:23:47 +01:00
David Given
c187b79d80 Add a 300RPM clock on 3[0] and a 360RPM clock on 3[1], for use with faking 
index pulses to the drive.
 2019-12-12 20:34:44 +01:00
David Given
072a097003 Archival (non-functioning) checkin of Verilog-based sampler code. Sadly, we've 
run into size limits for the device, and I need to slim down.
 2019-12-11 22:51:27 +01:00
David Given
a66e704bab Start ripping out the awful UDB-based sampler code, replacing it with a Verilog 
one and a standalone FIFO. This gets the FIFO working.
 2019-12-11 21:13:57 +01:00
David Given
32bb956710 Detect voltage levels *correctly*. 2019-12-11 00:05:34 +01:00
David Given
f436d6b582 Add a feature where we can measure the FDD bus signal voltages using the PSoC's 
ADCs. Increase the track step pulse width to 6us, because.
 2019-12-10 22:36:18 +01:00
David Given
91d6e9aeb9 Rewrite the sequencer engine with a separate fifo component and a pure verilog 
sequencer --- much easier to understand. We can write disks again!
 2019-11-25 20:52:13 +01:00
David Given
a40b26ff46 Archival checkin for trying to figure out why writes no longer work. 2019-11-24 15:14:32 +01:00
David Given
c266779433 Fix a bug where index pulses where being turned into flux pulses on read, 
leading to completely broken data whenever an index pulse happened.
 2019-08-27 23:58:07 +02:00
David Given
4954d33307 Add documentation for using the precompiled firmware. 2019-08-15 22:19:30 +02:00
David Given
56a36072f7 Sampler state machine cleanup; more debugging tools for the logic analyser. 2019-07-12 21:09:53 +02:00
David Given
1e3581c5f3 Turns out I was using the wrong error threshold flags for Mac disks. ND disks 
work fine now.
 2019-07-10 23:55:16 +02:00
David Given
c27c4fe312 Synchronise input pins and set LVTTL levels. More correct, but doesn't help the 
read issues.
 2019-07-10 19:48:03 +02:00
David Given
ad295c683c The sampler pulse conversion now uses BUS_CLK as the sample clock. 2019-07-09 22:43:46 +02:00
David Given
98ea5e9600 Connect up pins 2.3, 2.4 and 2.5 for debugging with a logic analyser. 2019-07-08 01:08:05 +02:00
David Given
ce6077fa22 Apply Denis Kushch's timing fixes to the schematic. No more warnings on builds! 2019-07-08 00:52:59 +02:00
David Given
c21177e2aa Finally make things work in release mode! 2019-05-01 13:06:43 +02:00
David Given
da8cae61b7 Add a debug UART back on pin 2.0 to get some semblance of debug information off 
the board. Discover a fun edge case where output transfers that were an exact
number of frames weren't being terminated correctly.
 2019-04-06 17:59:23 +02:00
David Given
bcc5a5f2cd Interim but working support for crunched data streams when reading from the 
device; writes haven't been converted yet. Reduces the bandiwidth from about
800kB/s to about 500kB/s, which is about what I thought.
 2019-03-26 23:03:19 +01:00
David Given
1d6f112b53 Remove the old logic-based sequencer. 2019-02-28 21:29:01 +01:00
David Given
f20ad8368a Non-working basis of a UDB datapath based sequencer. No more clock errors. 2019-02-28 01:03:58 +01:00
David Given
08cb20fd1d Remove the old, broken, Verilog sampler. 2019-02-27 23:57:13 +01:00
David Given
6bcd9a6644 The UDB-based pulse generator is back. The problem was I was trying to write to 
a HD disk (also the logic was wrong). I think my new Sony drive doesn't
autoselect.
 2019-02-27 23:34:55 +01:00
David Given
b8138aef40 Replace the UDB based pulse generator with the old counter-based one, which 
seems to work.
 2019-02-27 23:18:19 +01:00
David Given
462d15369e Create a pulse generator UDB component --- hopefully cheaper than what we're currently doing. 2019-02-26 23:33:06 +01:00
David Given
039d6b5fdd More fiddling with the UDB sampler. Still doesn't work. 2019-02-26 22:37:45 +01:00
David Given
74c6c6db56 Non-working attempt at replacing the sampler with a UDB datapath (so I can use 
the FIFO).
 2019-02-24 14:50:33 +01:00
David Given
34a4c888b5 Rewrite the Verilog *again*, and increase the CPU clock rate --- it's now a 
little more reliable at reading long intervals off disk, but does seem to be
doing it. I'm going to definitely need some better sampling logic here and
probably a FIFO. But it'll do for now.
 2019-02-23 19:51:27 +01:00
David Given
753a4a21ef Rewrite the sampler Verilog to use much more consistent logic. Still doesn't 
work for long intervals, though.
 2019-02-23 12:42:42 +01:00
David Given
6a1327e3f6 Attempt to translate the sequencer into Verilog and have it handle long 
intervals properly.
 2019-02-23 00:50:36 +01:00
David Given
970668aca5 Non-working attempt to replace the playback logic with Verilog. 2019-02-22 21:26:54 +01:00
David Given
a77925fe64 Replace the complicated sampler with 10 lines of Verilog; the hardware should 
now be able to handle any length of transition. And it's simpler.
 2019-02-22 20:01:34 +01:00
David Given
2527ac2ce9 Initial support for twin drives and 5.25" drives. 2019-02-11 23:43:44 +01:00
David Given
65315f4603 Take out the UART --- it's too much hassle. 2019-02-09 23:02:05 +01:00
David Given
e6a424d17a Abruptly realise that the capture logic was completely missing transitions too 
big to time; so, redesign it. Now the longest transition is 0xfe ticks and a
pulse is inserted if it's too long. Capture is way more robust now and the
resulting pulsetrain is a solid 198ms; and I wrote my first readable disk! (A
PC HD disk. Sadly, writing to the Brother didn't work.)
 2018-10-17 00:26:33 +02:00
David Given
99e222d012 Change the record mechanism to use an upcounter, not a timer. 2018-10-13 16:54:51 +02:00
David Given
aa7d420754 Start work on an autocalibrator to detect precompensation, but something's gone 
wrong and our reads are all bunk.
 2018-10-13 12:04:37 +02:00
David Given
b46d55cc07 The USB capture/replay interface now uses timestamps rather than intervals; 
this makes the hardware considerably simpler and more reliable (as I don't need
to spend time resetting the timers between pulses). Still doesn't help writes,
though. Simplify and improve clock detection; add an abortive attempt at an FM
decoder (turns out that the Brother doesn't use FM).
 2018-10-13 01:19:17 +02:00