Hi i’m just going to throw this out there. I’m spending a lot of time every day trying to build up my knowledge on theory and applications of ADC/DAC biz, both studying theory, taking notes on the various designs out there and also i’m building an ADC rn from someone elses schematics for getting control voltages in and out of the raspberry pi. My long term goals with this subject are
id like to have my own schematics and designs for ADC and DAC on hand that i can tweak for the tools i am creating because at a certain point i would like to stop focusing on nanocomputer boards for development and move to FPGA as my standard method because the nanocomputer environment makes certain aspects of high level coding easier but at the expense of really wasting a lot of the resources available maintaining the high level environment and being pretty suboptimal for any kind of situation where i want to be processing/generating simultaneous streams of data in synchronization. and also i want to just have some various of my own designs on hand that i can just etch to my own pcbs so that i can be as self sufficient as possible in this whole techno zone. and then of course open source all the stuffs i come up. id also like to understand electricty a bit better because i really don’t rn and at this point like eric p dollard makes as much sense to me as any textbook thing i read http://ericpdollard.com/free-videos/history-theory-of-electricity/
a side effect of this that might also be kind of handy would be to have an open source design for grabbing video signals from the external world ->adc-> encoding into a UVC signal for USB 3.0 for potentially a more flexible and higher quality method for getting analog signals into digital zones than the EZ cap family and also less bulky on the encoding side than the blackmagics. ( i haven’t really played with an elgato capture device, i have a hunch tho that because they are aimed at retro gamers that they might be working on a 360x240p sampling with line doubling vs sampling at 720x480i. this hunch is based entirely upon the powers of my imagination and the price point and consumer base of the elgatos and may have no actual relevance to any of the registry of possible universes). this end of things would also require a lot of studying on Encoding and Digital communications which i’m also pretty heavily into
big fan of these two lecture series Lec 1 | MIT 6.450 Principles of Digital Communications I, Fall 2006 - YouTube - YouTube
In a more abstract zone I think that there is an incredibly fertile and unexplored zone out there for hybrid analog/digital processing units. I am also a big fan of studying nonlinear dynamics, algorithmic complexity theory, and computability from an abstract viewpoint. From my experience working with signals it is obvious that there are really useful things you can do with analog voltage and analog photons that you can’t usefully emulate with dsp in any kind of high speed situation and vice versa. so i just want to have a playground for my own hybrid a/d processing unit so that i can experiment more with abstract signal side feedback systems that can utilize the strengths of electrons and photons for high speed iterative calculations. any kind of signal system where you are involving camera feedback and digital processing is already a hybrid analog digital computer! its just kind of too bulky most of the times to be working with like usbs cameras and operating systems and proprietary driver sampler/encoders gunking up the whole flow
if yr interested in like maybe joining in, having little meetups and shares about these ideas lmk! or if like you would just enjoy it if i like keep writing about what i’m learning or whatever just as a bystander for now also lmk!
and also if study groups ever seem like something that just isn’t a priority for this group also lmk and we can ditch this category too to keep things streamlined
Recently got more interested into analog video decoding/encoding, mainly cause I was looking for a good way to do YC separation. This led me to Analog Devices ICs, for example this evaluation board features a decoding and encoding chip, and the parameters are set using i2c.
ADC/DAC are integrated so I don’t know how much you can act on them, but there is a good list of parameters that can be set for both conversion, so that might be worth checking out. @cyberboy666 told me the Pi Capture thing was using a similar chip for decoding.
I’m currently doing a small board with the decoding/encoding chip, the idea was more of a analog converter to get proper YC out of composite, but since the signal is converted to digital and back, I was thinking to have the output of decoding chip and input of encoding chip exposed, so I can connect them with jumpers for my converter purpose, but it could also be tied to a pi or FPGA for processing, and then also converted back to analog.
Haven’t figured out how I will “talk” to it in i2c yet so I’ll probably also have it accessible with a connector so I can try with an arduino at first.
that looks like an interesting board to play around with! it looks like the digital output of the decoding chip is exposed on the pins already no? but if you wanted to feed it back into being encoded then you’d need to jury rig something?
i was looking at building my own adcs because it looks like flash adcs seem fairly easy to rig up and are incredibly fast, tho they do not scale well in terms of increasing bit resolution. but once I have some flash adcs to work with then i could use those in a pipeline format to experiment with how bit depths affect latency in that architecture. part of my motivation for this is to have an adc that can work with ‘glitch’ signals better as well, no off the shelf cvbs adc/encoder combo is going to do anything other than clamp out or reject out of bounds video signals so might be cool to try experimenting with something that is more amenable to ‘illegal’ signals
Yes the eval board has outputs for the decoder but the inputs of the encoder are hardwired to the decoder, so it would require cutting the 8 traces, and considering the price of the board, I’d rather make my own lol (I guess the price is also justified by the software they provide to edit the chips registers through usb).
Oh right, you were more thinking of “discrete”/flash ADCs, I guess if you have fast enough comparators you should be able to sample the whole CVBS signal with enough definition and then do the filtering/decoding in RGB colorspace digitally, as this is what those chips does (datasheet of the ADV7182 states about 57MHz sampling rate for CVBS).
I’ve tried analog ways to do YC filtering but it’s not really worth it as it’s either not so great quality or a bit too complex, sometime involving obsolete components, it seems that digital decoding is the way to go. I recommend this application note that review the various way to do YC decoding, might be helpful to check how it’s done digitally.
Then YC to RGB colorspace is done by amplitude/phase comparison of the colorburst with the active part of chroma.
But yeah all this is if you start with a CVBS/YC input, it would be simpler with component/RGB video to begin with, might just require more ADCs.
And of course if you want to have a kind of “flexible” TBC, you’ll have more control if you write the code/design the hardware yourself, as those decoder chip have a built-in “mini-TBC” (that’s what’s written in the datasheet lol), which is just some sync signals filtering/PLLing that can be turned on and off, not sure how it’ll react to glitchy signals (probably not really well).
I guess it needs some kind of frame/line buffering to achieve a proper correction, or at least freeze on the last intelligible line/frame.
I only know really basic code so that’s why I was more interested in ICs with integrated decoding, but as long as you’ve sampled your signal(s), everything else can be done digitally and you’ll have more control on every step of the process, that’s for sure
yeah i was definitely not thinking about doing that much decoding on the analog side of things! the general thing that turns me off of the idea of delving too heavily into the on the analog signal processing side of things is the general expense of the components and the fact that from my end of things as someone who’s worked a lot with code it seems like a lot of my goals are stuffs that is just handled better in dsp zones. but i also want to have a lil laboratory set up too where i can try certain things both in the analog side of things and in the digital side of things and be able to like actually compare to find out for myself
