exxos's DFB1 trials

[Badwolf]

On the scope.. 10ms /timebase

Here's my 5V header output.

Same 'low', but it's not really visible.

5V.png (7.15 KiB) Viewed 1299 times

Will see what I can do about dropping the voltage.

BW

[Badwolf]

Does the FPU actually heat up running that or is it quite happy?

Driving mine at 30 MHz (the highest that passes all tests), after about 2hr45 it's cooler than the CPU.

My highly calibrated finger reckons it's about 50C to the CPU's 60.

BW

[dml]

@dml, if I know anything about fractals, I suspect each render has a completely different draw time on account of it varying what is being shown, but I do wonder if I made a mistake earlier and quoted a 50/16 minutes-per-frame figure of 50.

It probably does have a variable render time yes.

It is not ideal as a benchmark but it does have a long runtime and crashes if the FPU is not stable, so it has some value as a burn-in test.

(I don't have the source anymore so I can't improve on its benchmark properties - but I might be tempted to do another one when have a bit of time).

[Badwolf]

On the scope.. 10ms /timebase

Here's my 5V header output.

Same 'low', but it's not really visible.

5V.png


Will see what I can do about dropping the voltage.

BW

@exxos

My ATX power supply give me 4.44V. CPU mostly works (had one hang), but the FPU, at it 30MHz limit, starts failing regularly now -- the trig functions first to go again.

We could be onto something here.

BW

[exxos]

@Badwolf Your 5v looks a bit more stable than mine but with the things spiking anyway its hard to see if rail noise is a factor.

Interesting your failing on the FPU with the atx. Be interesting if you do the same experiment with the one I'm sending. The FPU runs at 40mhz and the CPU 40mhz. But not the same osc. The cpu just won't run at 50mhz. So I guess you can try with your falcon PSU and atx at 50mhz.

I'll have to hookup one of my PSUs in the falcon when I get chance. I can alter the 5v on the new ones.

It is a little worrying though as I think we both have a recapped PSU. I bet half the people who would want DFB1 won't have a recapped one. So things could be even more unstable for them. We may be able to filter some noise but not much we can do about sagging voltages. The falcon PSU doesn't have a adjuster for the 5v.. Though I guess I could fiddle that ... Problem is there isn't a schematic of the falcon PSU But I guess I can follow the 5v feedback circuit. I can't imagine its much different than any other PSU.

[exxos]

Wasn't sure if to put this here or in my blog.. But thought I would dump it here has this is the thread we was talking about voltages on the power supply..

IMG_0293.JPG (177.66 KiB) Viewed 1234 times

IMG_0292.JPG (229.4 KiB) Viewed 1234 times

So I reversed engineered the feedback network saw something rather interesting...

1.jpg (35.65 KiB) Viewed 1234 times

It is a little unusual to have a exact 2.5V resistor divider output. Normally they are biased to more like 1V.. but anyway.. The seemingly pointless 330K resistors which are also in parallel are what interested me.

Because basically I was going to do the same. Slap something like 100K across the feedback resistors to increase or decrease the voltage.. And yet there is already 330K on there! so basically if I simply cut one out the voltage will go up or lower slightly!

So as I want the voltage to increase slightly, I will chop out the top 330K (R19 ) and see what happens...

[exxos]

Well that was hardly worth the effort then It was 4.88V before.

IMG_0301.JPG (236.55 KiB) Viewed 1223 times

IMG_0300.JPG (35.71 KiB) Viewed 1223 times

[exxos]

Going back to my original plan of just rolling the dice and slapping 100K on the thing...

IMG_0303.JPG (239.6 KiB) Viewed 1219 times

IMG_0304.JPG (31.09 KiB) Viewed 1219 times

So I guess technically 47K should up it to around 5.1V or thereabouts.

Problem being now is that I sent the booster back to @Badwolf

[Steve]

On a side note, while I was working on different people's Falcons doing re-caps etc, I did come across PSUs with different resistor combinations there with slightly different 5v outputs. Later I'll go through my old photos and see if I can find it.

Perhaps BW can take a photo of his because he gets about 5v doesn't he.

[Badwolf]

On a side note, while I was working on different people's Falcons doing re-caps etc, I did come across PSUs with different resistor combinations there with slightly different 5v outputs. Later I'll go through my old photos and see if I can find it.

Perhaps BW can take a photo of his because he gets about 5v doesn't he.

Interesting.

At work now, but I dont know if they're sufficiently visible in one of my old videos?

This is the main Falcon PSU one, but it's filmed in potato-vision.





BW