Bit Angle Modulation (BAM) in a PIC

[Mike, K8LH]

Is there any way I can get temporary access to locked post above to update the program listing? I posted incorrect version which has some bugs. Sorry!

[Mike, K8LH]

Darrel (and gang),

A long overdue assembly language Charlieplexed BAM example for 20 LEDs on a 12F683. Example uses 8-bit 256 step BAM driver with a 1-usec step size. Refresh rate is 781-Hz with a 1280-usec frame rate. LED "0" is animated. That is, it ramps from duty cycle value of 0 up to 127 and then back down to 0. LED "10" through LED "19" are lighted with duty cycle values of 1 through 10 so you can check out "low end" illumination.

You might just hook up a single LED for LED "0" to check out the program. Just connect the LED cathode to GP5 and the LED anode to GP0.

Kind regards, Mike

[Mike, K8LH]

Has anyone tried "gamma" correction to smooth out LED brightness level changes? I just added a 64 level rom "gamma" array to the program in the previous post and it made an amazing difference in the smoothness of the fade.

[Ioannis]

I supose this depends on the type of LED. Each type of LED may need its own gamma curve.

Ioannis

[Mike, K8LH]

Bummer! I was afraid of that...

[Sneaky-geek]

Hi Mike,
Fantastic work!

Could you please post:
LED Part number,
Gamma correction routine?

Please, Please, Pretty Please?

I've learned a lot from this thread and want to say Thank You, to ALL who have contributed.

73,
Terry K9HA

[Mike, K8LH]

Hi Terry, K9HA,

Attached is assembler source with "gamma" table. I also slowed down LED "0" fade loop time considerably to look for "stair stepping" effect since I was only using 64 "gamma" steps but the fading is remarkably smooth.

I'm using Fairchild MV57164 Bar/Dot displays but almost any LED should work fine.

A couple observations;

(1) I'm very surprised at the light produced from LED "10" which is set to a duty cycle of 1 (1-usec -> 1/1280th or 0.078125% duty cycle).
(2) Mirrored BAM doesn't seem to be necessary to correct the 127-128 duty cycle transition visual artifact when using a multiplexed display (with gaps between LED updates).

Regards and "very 73", Mike, K8LH

[sanch0]

Has anyone tried "gamma" correction to smooth out LED brightness level changes? I just added a 64 level rom "gamma" array to the program in the previous post and it made an amazing difference in the smoothness of the fade.

yep... tryed it in 3 ways:
1. Simple ISR-based software PWM with floating (table-generated) pulse widths.
2. Strobing /OE of external latches (54 PWM channels/PIC16f877a) with HW PWM output (table-generated) + SW PWM (i had to slightly increase LED current to get more brightness, as the full duty cycle hadn't been achievable, almost like with 1.5x charlieplexing)
3. Increased bit depth to 13(14) bit and PWM speed (refresh rate) to 610Hz on 18F for better perfomance.

[sanch0]

2 ALL...
There is one small problem with high speed BAM/MIBAM. Problem exists if u use simpler MOSFET driving scheme (or if u use slower bipolar schemes like darlinghton's transistor pairs and if actual switches are controlled by MCU's outputs directly) or if u need higher current per channel, especially if currents on some channels are significantly different. This happens because switching delays differ too. Check out datasheets for gate driving characteristics for higher current applications and compare it with needed signal discretization. PWM methods don't solve the problem but it need less signal transitions per cycle and work slightly better for lower bit depths without additional gate drivers (for reasonable currents, of course). Actually this problem just increase duty_cycle/brightness relation's non-linearity, but combined with longer wires, bad PCB layouts etc. it may cause unpredictable results for visual perception. So don't be surprised). sorry, can't describe problem in detail because of my poor english (lack of practics) and lazyness. Anyway, best solution is to check your application's requirements in advance))).

[lanyong]

hi,sanch0.

how r u?

i am waiting for ur codes for 16 channels of 14-bit BAM, at 610Hz.

where are them?

thank u.

[sanch0]

hi,sanch0.

how r u?

i am waiting for ur codes for 16 channels of 14-bit BAM, at 610Hz.

where are them?

thank u.

done and working already... for now my code gives 16 bit (E)BAM at 610hz (actually 14bit at 610 hz (1 Tcy discretisation) + 2 bits at 610/2 and 610/4 Hz and channel values are updated at 610/4Hz) my code gives real 610hz refresh rate because i don't use MIBAM, which would steal additional frame giving in result twice lower framerate (for visual perception). for now 16 channel code is working (with same 28-pin PIC18xxxx PCB, i've posted it's photo before), more channels may be easily added if needed (it's possible to make up to 64 channels with external latches upto my calculations, more with internal latches). Actually my algo has some more improvements compared to MIBAM. but i'm slightly drunk for now to describe it)))...
Sorry guyz, i got no access to internet from home, checking mail very rarely. have no normal job for long time to pay for all stuff i need... Stupid country, stupid government, stupid myself)
best regardzzz