• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Calculating Video Bandwidth/Signal Frequency

Joined
Aug 17, 2008
Messages
2,188 (0.53/day)
Location
Minnesota, USA
System Name TaichiTig
Processor 6800K @ 4.2GHz
Motherboard ASRock X99 Taichi
Memory 32GB DDR4 3200
Video Card(s) 1060 3GB
Storage 1TB 960 Evo, 1TB 850 Evo, 7x 6TB X300 in DrivePool
Display(s) BenQ PD3200U + Samsung C32HG70
Case Antec Twelve Hundred
Audio Device(s) Behringer UMC404HD, 2x JBL LSR308, Sennheiser HD558, Beyerdynamic DT-1350, Etyomic HF3
Power Supply Corsair 750TX
Mouse Logitech G502
Keyboard Deck Legend
Software Win10
Trying to clear up a number of things here.

I want to determine the exact meaning of the listed bandwidth of products like this (lists 400MHz bandwidth/2048x1536 @ 72Hz). Other products (example) list a "pixel frequency". Still elsewhere I've seen "pixel bandwidth" and "signal frequency".

So firstly, are the terms "pixel bandwidth", "pixel frequency", "signal bandwidth", and "signal frequency" interchangeable in the context of analog video signal transmission?

Intuitively, I thought video signal bandwidth would be simply pixels multiplied by refresh rate. This page, however, says the actual calculation is [(pixels * refresh rate)/2]3.

Thus, the signal frequency of 72Hz 2048 x 1536 video would be calculated as:

[(2048 * 1536 * 72)/2]3 = 340MHz

....which is closer to the listed 400MHz than the calculation I intuitively expected:

2048 * 1536 * 72 = 227MHz

....but still significantly off.

So secondly, where does the extra 1.5x factor (result/2*3 from this page again) come from?

Thirdly, why does it still not match the numbers provided for the first product I linked to? Is it just that 2048x1536 @ 72Hz is the highest "relatively standard" display configuration that doesn't exceed 400MHz? Or are they actually using a different calculation? Because the way I'm calculating it, 400MHz could handle 2048x1536 @ 84Hz.

I just want a solid understanding of video signal bandwidth as I frequently use custom resolutions and refresh rates, so want to fully understand what these products can handle.
 

Easy Rhino

Linux Advocate
Staff member
Joined
Nov 13, 2006
Messages
14,784 (3.11/day)
Location
Mid-Atlantic
System Name Desktop
Processor i5 7500 Kaby Lake
Motherboard AsRock MicroATX Z170M Extreme4
Cooling Stock
Memory 2x 16G HyperX DDR4 2667
Video Card(s) AMD Radeon 5700
Storage Samsung NVMe 512
Display(s) LG 34GK950F-B 34" 21:9 Ultragear WQHD Nano IPS Curved Gaming Monitor with Radeon FreeSync 2
Case Corsair Carbide 400C
Audio Device(s) On Board
Power Supply Seasonic SSR-650GD Flagship Prime Series 650W Gold
Mouse Logitech G900 Chaos Spectrum
Keyboard Code V2B
Software Windows 10
400MHz is most likely simple marketing. my guess is that there is no standard refresh rate between 72hz and 120hz. so obviously a 120hz refresh rate would significantly reduce resolution is could handle and it most likely is not meant to handle that high refresh rate in the first place.
 

Easy Rhino

Linux Advocate
Staff member
Joined
Nov 13, 2006
Messages
14,784 (3.11/day)
Location
Mid-Atlantic
System Name Desktop
Processor i5 7500 Kaby Lake
Motherboard AsRock MicroATX Z170M Extreme4
Cooling Stock
Memory 2x 16G HyperX DDR4 2667
Video Card(s) AMD Radeon 5700
Storage Samsung NVMe 512
Display(s) LG 34GK950F-B 34" 21:9 Ultragear WQHD Nano IPS Curved Gaming Monitor with Radeon FreeSync 2
Case Corsair Carbide 400C
Audio Device(s) On Board
Power Supply Seasonic SSR-650GD Flagship Prime Series 650W Gold
Mouse Logitech G900 Chaos Spectrum
Keyboard Code V2B
Software Windows 10
hrm, that is confusing. does this help?

http://www.csgnetwork.com/videosignalcalc.html

my guess is that since the calculations are considered conservative a more reasonable number would be closer to but not higher than 400MHz when you take into account filtering and noise? again, just a guess.
 
Last edited:

Easy Rhino

Linux Advocate
Staff member
Joined
Nov 13, 2006
Messages
14,784 (3.11/day)
Location
Mid-Atlantic
System Name Desktop
Processor i5 7500 Kaby Lake
Motherboard AsRock MicroATX Z170M Extreme4
Cooling Stock
Memory 2x 16G HyperX DDR4 2667
Video Card(s) AMD Radeon 5700
Storage Samsung NVMe 512
Display(s) LG 34GK950F-B 34" 21:9 Ultragear WQHD Nano IPS Curved Gaming Monitor with Radeon FreeSync 2
Case Corsair Carbide 400C
Audio Device(s) On Board
Power Supply Seasonic SSR-650GD Flagship Prime Series 650W Gold
Mouse Logitech G900 Chaos Spectrum
Keyboard Code V2B
Software Windows 10
It looks like that is using the formula with the 1.5x factor. The fact that it also mentions that it's using a "Sony formula" and that it is conservative makes it particularly useful to me. Am going to look into that but sounds like there's just plain no true industry standard for this.

I'll do some more searching and try contacting a few companies to find out what their methods of calculating it are. Will post if I get useful info.

Thanks very much Rhino.
yea report back. that is interesting since i dont have much knowledge in the field of electrical engineering.
 
Joined
Aug 17, 2008
Messages
2,188 (0.53/day)
Location
Minnesota, USA
System Name TaichiTig
Processor 6800K @ 4.2GHz
Motherboard ASRock X99 Taichi
Memory 32GB DDR4 3200
Video Card(s) 1060 3GB
Storage 1TB 960 Evo, 1TB 850 Evo, 7x 6TB X300 in DrivePool
Display(s) BenQ PD3200U + Samsung C32HG70
Case Antec Twelve Hundred
Audio Device(s) Behringer UMC404HD, 2x JBL LSR308, Sennheiser HD558, Beyerdynamic DT-1350, Etyomic HF3
Power Supply Corsair 750TX
Mouse Logitech G502
Keyboard Deck Legend
Software Win10
Reporting back. A user at OCForums (where I also posted this query) posted the following response yesterday (link):

nakana at OCForums said:
I'll take a stab at this. So we want to know where Extron is getting their 1.5x factor, or divide by 2 and multiply by 3.

We'll start with where dividing by 2 comes from. If you multiply horizontal pixels by vertical pixels by the refresh rate and call that the bandwidth. You are saying that that is the Hz required (wavelengths per second). This assumes that each pixel requires a full wavelength to be represented. Let's consider an image with each pixel alternating: black/white/black/white/.... over the entire screen. Does a black pixel require a wavelength (a voltage change) to be represented? No. A black pixel is a flat low voltage, and a white pixel is a flat high voltage. Therefore one wavelength is created by 2 pixels, and it looks like a square wave. There's the divide by 2.

As for multiply by 3, it would be tempting to think that you need to multiply by the number of color channels (red, blue and green), but since each channel is delivered on a separate cable this is a red herring.

In truth the multiply by 3 is a safety factor. A fixed length of coaxial cable will perform very well at low frequencies, but at extremely high frequencies it will attenuate the signal. This attenuation is created by the effect of cable capacitance on the alternating current of the signal. A technical name for this is an RC circuit or filter. You may know it as a low pass filter. Pictured here:

.

You can see the response is pretty even until it begins attenuating past a certain point then it starts to attenuate quickly. -3dB attenuation is considered the greatest permissible amount of attenuation. This is equivalent to half power. Someone could argue that "based on real world viewing tests it's fine for the signal to be attenuated more" or the opposite - that "viewers will notice degradation at -2dB". Either way it's an accepted standard (according to Infocomm).

I'll consider an example with XGA, 1024x768, at 60Hz.

1024*768*60 = 47185920 pixels per second.
47185920 / 2 = 23592960 Hz (as discussed above)
23592960 Hz / 1024 / 1024 = 22.5 MHz

If you used a coax cable rated to -3dB at 22.5 MHz there would be -3dB signal attenuation at 22.5 MHz. To keep the entire signal in a safe, low-attenuation, range you multiply by 3. It's just a rule-of-thumb convention. Multiplying by 2 would be OK. Multiplying by 4 would be playing it unnecessarily safe.

This also explains why there can be variance like what you're seeing with the 400MHz (-3dB at 400MHz) product that is advertised for 2048x1536 @ 72Hz. It never hurts to have a little extra headroom.

Everything I've said above only applies to uncompressed analog video signals on coaxial cables.

Let me know if anything is unclear or if I created new questions.
Generally I wouldn't revive an old thread like this but it's a good explanation for anyone else who comes searching for answers in the future.
 

Aquinus

Resident Wat-man
Joined
Jan 28, 2012
Messages
11,459 (4.01/day)
Location
Concord, NH
System Name Kratos
Processor Intel Core i7 3930k @ 4.75Ghz
Motherboard ASUS P9X79 Deluxe
Cooling Corsair H100i V2
Memory G.Skill DDR3-2133, 16gb (4x4gb) @ 9-11-10-28-108-1T 1.65v
Video Card(s) Sapphire AMD Radeon RX Vega 64
Storage 2x120Gb SATA3 SSD Raid-0, 4x1Tb RAID-5, 1x500GB, 1x512GB Samsung 960 Pro NVMe
Display(s) 1x LG 27UD69P (4k), 2x Dell S2340M (1080p)
Case Antec 1200
Audio Device(s) Onboard Realtek® ALC898 8-Channel High Definition Audio
Power Supply Seasonic 1000-watt 80 PLUS Platinum
Mouse Logitech G602
Keyboard Rosewill RK-9100
Software Ubuntu 18.04 (5.3.0 Mainline Kernel)
Benchmark Scores Benchmarks aren't everything.
400Mhz is the maximum frequency of the DAC on the video card. Consider for a moment that VGA (that is what you're talking about,) uses an analog signal and the card needs something to make that signal, which the DAC is used for. Not running it at 100% most likely is more to keep the DAC within safe limits within the realm of what it can do which is why you won't see it get all the way to 400mhz output. All in all, as video cards stop supporting VGA I think we will see video cards start ditching the DAC all together which I can't wait for. :p

Edit: Just noticed that this is an old thread. WHY YOU REVIVE IT?! :eek:
 
Top