PlanetZ forums - scopeusers.com

I've a trouble that afflicts my system.
I've just bought a Pulsar2 to add dsp power and new devices to my Pulsar1. I've attached the stdm cable and installed the drivers and the devices. I've tested it and everything works correclty until I set a samplerate of 96kHz.
This appears:


I think to have installed correclty the cable, but seen this error i'm not totally sure. Anyway the dsps seem to work correctly :


Any ideas? Suggestions? I forgot something?

Many thanks

Were you able to run 96kHz with your Pulsar1 alone, before?

In DSP Meter, is the Pulsar2 listed as first? The list should read 1/1,...,1/6,1/4,...,4/4 then you know the 6 dsp card is first. This is the card SFP will talk to, and that will be 'master' of the chain.

the list above reads 1/1...1/6
looks like you switched a fairly 'good' equipped project to 96k, which produced an overload with the false cable message.

start with a less crowded one before you switch to 96k or start from a default one at the high sample rate.
there are some (well...) constraints for 96k, specially if a first generation board is involved.

cheers, Tom

Hi Atom,
Yes, I was able to run my pulsar1 at 96kHz before.
I don't know if I understand correclty what you are asking to me (ahhh, my poor english...), but yes, I've set the pulsar 2 as the first of the list. So I think it's master also.

Are you thinking about a problem with the pulsar2?

Watching the connectors, I can see 2 slots for the stdm cable on the P2; can be possible that I've choosed the wrong one? Or can be possible that I've mounted the stdm with the wrong orientation?

Hi Tom,
The problem appears with any project. I've noticed it for the first time with my standard progect wich is basic: only a single channel mixer, wave source and audio dest. (less than 1 dsp used).

The one I've used for the screenshot, was voluntarily overcrowded, because I want to show that all the dsps are usable at the same time, avoiding any "hey you forget to attach the stdm cable" misunderstanding

On 2005-07-12 04:51, Lima wrote:
... was voluntarily overcrowded, because I want to show that all the dsps are usable at the same time, ...

got me there...
have you modified latency ?
the cards might run in a 'wrong time grid', so even reducing it could improve the situation - or the Pulsar One (slower IO) misses a signal (pure speculation, tho)

cheers, Tom

At 96 khz your DSP will be loaded the double of how much they are at 48 khz. So you can't expect to go higher than 45-50% at 48 khz and then be able to switch at 96 khz.

Thanks guys for your help! It's very appreciated, indeed! I hope to find the way to resolve this.

Well, about the latency, I can say that the first thing I've done as soon as I've mounted the 2 boards is to set it to 3ms @44100Hz
I was pretty sure that, when I would have exceded the 6 dsp usage, any kind of error would be occured to warning me that I can't mantain that latency.
Instead of this, nothing appears when I go use more than 6 dsp with 3ms of latency. When I see this I've thinked that probably the secondary board will share only the Dsp power, and maybe this can be done without any problem of latency(anyway this sounds to me quite strange...)

Despite this, the error occurs also for long latencies, like this screenshot shows:


To Alfonso: this shows that it occurs also for simple projects like the one used here
Just to be regorous the operation I've done was:
Close scope, reopen it with my standard project (wich is this one @44100 Hz and 34 ms latency), re open this project and switch the latency to 96kHz.

Well doctors, is the patient serious?

looks bad.
luckily, the cards are fixable. have you wiggled the whip gently to see if there is a connector problem on the card?...unlikey to help, but....

The dsp load is not affected by the lower latency, but the computer performance is. There are different concurrent effects when you double freq.rate. The 3ms at 48 become 1ms at 96, PCI bus flaws could happen in average power machine. So, higher freq.and lower latency both independently put a load on computer resources.

What happens if you set ULLI at 7ms or 13 ms? because if you can then switch at 96khz without hassle your latency will change authomatically to 3 or 6 ms's, that's very workable and you will have the proof that it's only a matter of computer power. I don't know if an average machine could really handle 1ms of latency.....

Hi Garyb,
I've connected the cable quite gently. But when I've connected the 2 cards I havn't heard the classic "click"; but I push with a reasonable force in my opinion.
Does your board have make that click?

If I pull the cable (obviously when my pc is shutted down) It come out without any particular opponent force. Sould I trust myself and push more?
I'm worried to make a damage pushing more, what you suggest?

To Alfonso, I've just tested all the ULLI possibilities, but no one works @ 96k...

It's possible that I've mounted the cable in the opposite direction?
Anyone knows the effects of mounting the cable wrongly?

The STDM cable can be mounted in any direction.
It's recommended, if you have 2 cards, to connect the 2 extreme connectors of the cable, leaving empty the one in the middle.

I think that if everything works ok at 48 khz and the problem arises only at 96 khz it shouldn't depend from the cards or the cable,otherwise they wouldn't work at all, but more probably from computer performance issues, due to IRQ issues, pci bandwidth, bottlenecks or whatever is pushed hard by the change of freq.

If you stay at 44.1 or 48, how many Masterverbs can you load before you get pci error messages?

I've done the test. At 48 kHz I can load 9 Masterverb. How do you think?

ops, stange: closing the project I've got either a dsp overload error (with optimization I've resolved it) and a PCI overload.
Why, seeing that before it works correctly?

Lima, running such a setup you should also take into account some DSP organisation, and that STDM can carry 256 44.1 channels AFAIK.

- Use IO where they fit best: if you route ASIO to ADAT output, choose the ADAT's on the primary card: this can save 16 STDM channels.
- Use small mixers, try putting them on the other card: you can lock most of them to a certain card, which goes where will be self-explicatory when you have the Routing Window in front of you. For example if 16 ASIO channels are mixed and output straight to analog, the mixer can be locked to the first card, and the card's output can be used, and putting effects inside the mixer will force them on that card too, saving STDM bandwidth while keeping an eye on the DSP Load Meter (keep it set to display % so it's readable for us humans).

From Pulsar2 on, the cards have multiple STDM connections. This STDM error won't happen between those cards, the extra cables can handle the required bandwidth better.

Hope this helps. If you're filling up DSP to benchmark it, you will always have this problem. If you can save 16 or 32 audio channels over STDM, your odds on a Scope session without headaches will be much better.

you may want to try switching the clock from master to slave wait and see if it locks and then switch it back.

also as others noted don't max out your DSP then select a higher sample rate.

you may want to try a differnt STDM cable
or clean the one you have. Also clean the Gold contact Where STDM connects.

under hardware I/O in the routing windows which card do you see first? is it the Pulsar or the Pulsar 2

<font size=-1>[ This Message was edited by: antar on 2005-07-14 13:10 ]</font>

On 2005-07-14 08:04, Lima wrote:
I've done the test. At 48 kHz I can load 9 Masterverb. How do you think?

mmm, not superb but not bad either, you don't show a bad pci bus....

Thanks my friend, now I try all the things you have suggested. I'll keep you informed,
I also try to change the bus of the pulsar2 where I've connected the stdm.

Stay tuned!
Thanks again

yes, that's important. see the install manual...