Yes, it would be very interesting indeed to record separate tones.
If I understand correctly, the tones never physically mix, so I think there's no problem keeping them at close frequencies, and making them both nominally 3kHz or 3.15kHz or thereabouts. Conserves existing calibration, and makes things standard. Also makes it easy to display on the same chart, in principle.
Thanks and all good stuff........!
LD
Last edited:
Ref. to the quotes issue, I am running the code on both a 64 and 32 bit computer. The former does not need quotes when entering the file name whist the latter does. It confused me for a while until it was mentioned in this thread. I am running the code from PyCharm which uses the Python interpreter, part of the Anaconda install, on both computers.
It’s not an issue but something to be aware of for a novice like me! Your mention of adding a note at the prompt would be worthwhile though.
It’s not an issue but something to be aware of for a novice like me! Your mention of adding a note at the prompt would be worthwhile though.
As a physicist, I can safely say that nothing is exactly my thing !
If I understand, the idea is to seperate record separate tone tracks, one from the record and one somehow synthesised from the platter ? So, to keep them identifiable, all we need to know is which is left and right channels in the stereo recording?
From my point of view, my SW is calibrated and well trodden for tones near 3kHz. Though it can, in principle, work with (almost) any carrier frequency, I haven't run calibration checks other than near 3kHz. And widely separated carrier frequencies might have different issues, as we found at 100Hz......
So both at the same nominal frequency and near 3kHz gets my vote. The python boys have this f calibrated too AFAIK. Though I can cope with most anything that the python boys prefer...........
LD
Easy if one is in each channel sinch it's a trivial change to read both channels. If we want to standardize on mono files are just what we have and stereo files use the cal it would work for everyone. I already test for stereo channels and use only the first.
Taking left for one rev and right for the second (as an experiment) was amazingly uninformative. Every file I tried they were right on top of each other, so having stereo capability has little other use. If the frequencies are different the display will get wonky trying to put the high frequency on the outer radius the lower could disappear into the origin.
Yes this is an issue in the Python community and you can find plenty of threads on it. I will have a new rev where no quotes (required) should work for anyone. I added a "no quotes" to the prompt.
I'm adding some variables at the top of the script so one change can set the RPM and the post filter analysis frequency (defaults will stay 33RPM and 100Hz filter). The filters are now linear phase so no difference from LD's data should show up even at 20Hz filtering. I ran cal files and the numbers were right on and the AM rejection was remarkable. The big difference is in how the true FM noise is removed, the Hilbert transform used on one whole rev at a time returns the FM noise all the way out to Fs/2 and this needs to be filtered to be useable.
If we can decide on a format for the FFT data I can work on that, though as I said averaging more revs could give more resolution. There may be other strategies to improve that data too.
Last edited:
Direct digital synthesizer, they are common in coms to generate arbitrary local oscillators often to create exact quadrature sine waves of arbitrary frequency.
Originally we talked about the W&F track being in one channel, and a momentary signal in the other to index the platter once per rev.
I'm taking about adding another index signal for the record itself. I'd use two IR tracking sensors or similar - one for the platter, and one for the record. That way when the record is rotated between captures we'll know its exact position relative to the platter across captures.
So, I've two momentary index signals I need to stick somewhere in two audio channels.
Ah, confused yoda is......
......I thought the DDS track is to be a report of instantaneous platter rotation speed, whereas that recorded from the record includes the artefacts of record/arm/cart ?
Much depends on the derivation of the DDS track, and I think to be useful it has to be a direct real time report of platter rotation speed - so hooking into the motor drive PLL, for example, wouldn't do this because we don't know the transfer function between motor drive PLL frequency and real time platter speed..............
???
LD
......I thought the DDS track is to be a report of instantaneous platter rotation speed, whereas that recorded from the record includes the artefacts of record/arm/cart ?
Much depends on the derivation of the DDS track, and I think to be useful it has to be a direct real time report of platter rotation speed - so hooking into the motor drive PLL, for example, wouldn't do this because we don't know the transfer function between motor drive PLL frequency and real time platter speed..............
???
LD
Ya'll are overthinking this.
We'll have the signal from the cartridge recorded to one channel to capture the W&F track.
We'll have an optical pick-up triggered by an index once per revolution of the platter. We'll have an optical pick-up triggered by an index once per revolution of the record. This will allow us to know the relative position of the test record to that platter. In turn we can change the record to platter relationship (e.g. rotate the record 90deg) between test track captures to determine if issues are with the 'table or the record.
The DDS is just because I need a way to capture the index signals in an audio channel, and I happen to have a module I created to replace an IC on Technics tables that has a DDS and a PIC MCU. A few lines of code and I'm up and running.
So the question is, with two index signals and one audio channel, how do the software guys want to tell them apart. Frequency, duration, or amplitude?
We'll have the signal from the cartridge recorded to one channel to capture the W&F track.
We'll have an optical pick-up triggered by an index once per revolution of the platter. We'll have an optical pick-up triggered by an index once per revolution of the record. This will allow us to know the relative position of the test record to that platter. In turn we can change the record to platter relationship (e.g. rotate the record 90deg) between test track captures to determine if issues are with the 'table or the record.
The DDS is just because I need a way to capture the index signals in an audio channel, and I happen to have a module I created to replace an IC on Technics tables that has a DDS and a PIC MCU. A few lines of code and I'm up and running.
So the question is, with two index signals and one audio channel, how do the software guys want to tell them apart. Frequency, duration, or amplitude?
I get it now, you want a tone that is interrupted for the index. Not sure that's needed, as the tape tab, or whatever, would make it's own blip, still the interrupted sine wave wouldn't hurt.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.