[Ben Krasnow] Hacks E-Paper For Fastest Refresh Rate

[Ben Krasnow] is known for his clear explanations alongside awesome hardware, being one of only a few hackers who owns an electron microscope. This time he’s explaining how E-paper works while modifying the firmware of a 4.2 inch E-paper module to get a higher refresh rate. As for the awesome hardware, he also analyses the signals going to the E-paper using an ultra-fancy loaner oscilloscope.

E-paper explanation diagram
E-paper explanation diagram

After starting out with a demo of the firmware in action before and after his modification, he explains how the E-paper works. The display is made up of many isolated chambers, each containing charged particles in a liquid. For example, the positive particles might be black and the negative might be white. By putting an electric field across each chamber, the white particles would be attracted to one end while the black would be attracted to the other, which could be the end you’re looking at. He also explains how it’s possible to get a third color by using different sized particles along with some extra manipulation of the electric field. And he talks about the issue of burn-in and how to avoid it.

Having given us that background, he then walks us through some of the firmware and shows how he modified it to make it faster, namely by researching various datasheets and subsequently modifying some look-up-tables.

Turning back to the hardware, he shows how he scratches out some traces so that he can attach scope probes. This alone seems like a notable achievement, though he points out that the conductive layer holds up well to his scratching. At that point he analyses the signals while running some demos.

The result is the very informative, interesting and entertaining video which you can watch below.

We have a lot more of these amazing adventures with [Ben] here on Hackaday. We really enjoyed seeing his entry for the flashing light prize where he heated a coin cell to squeeze out some extra current. And his 2016 Supercon talk on rapid prototyping is a pep talk for all hackers.

26 thoughts on “[Ben Krasnow] Hacks E-Paper For Fastest Refresh Rate

  1. Impressive!!! I’m not sure why he scratched the pixel traces? I was reading something else also while watching. Was that to determine their signal effect on the code? I’ll have to try to watch again later.

    Excellent attention to detail and I don’t use the e-paper though makes me think I can upgrade the old TDS-520 to run off Windows XP or 7 Pro and really make a more awesome custom hacked oscilloscope cause my budget can’t afford the one shown. Why not, even if planning for a few years out, I need to read up on the schematics, JTAG’s, test points and get over the GPIB hurdle then work on some other mods?

      1. Yes, macona no doubt. I wasn’t meaning I’d use the TDS-520 as the motherboard. I’m about to look into running off of 12-24VDC if worth the effort as an option to quiet up the system as a whole. I’m already detailing out the CRT to LCD swap. Since there is room on the inside, this got me thinking I can put more items in the case that is already shielded even if I put the motherboard and HD and interface card (if I don’t custom) in a shielded case. Would be using another desktop or laptop and maybe even a tablet as the motherboard since would emit less signals. A laptop would be noisy as well as a desktop. A tablet I’m thinking isn’t as noisy. I don’t think I’m going to try to put a battery pack in since to many much power requirement. DC will avoid running the inverter.

        1. I see, there is a company that sells a LCD retrofit for the 520. You could also tap into the VGA out if your scope has that option.

          I would not put anything else into the case, emissions from the extra components could interfere with the scope itself. Running it off of DC might be possible with a bunch of DC/DC converters but you will have to get an isolated DC/DC converter to supply the -15v rail.

          1. Yeah, I picked up an arcade CRT to LCD converter for an LCD screen I have that will fit and I also found a $5 Haier mini TV that will fit at the Goodwill store. Thought the later would be more humorous though the screen isn’t as tall, about 2″ shorter… probably will use for another project.

            A gentleman mentioned I may be able to directly connect to the VGA pinouts on the 520 board (or video board) and then I can split to the backside. Even with the conversion board, I’ll split out the back too.

            I was chatting with Tektronix earlier today about the software tekVISA and downloaded Wavestar_2012.05. I was referenced some links to two manuals and the Frankenstein Tablet controller for Tektronix USB spectrum analyzers using a Toughpad example as well as I’d have to call Tech Support for more details and given their number which I still have to call. I’m using NI PCI to GPIB and will most likely use the NI-VISA/488.2 software. Really awesome the ability to custom program applications with the libraries.

            I was thinking if I put the computer in a mu-Metal or shielded case I may be able to shield without adverse impact to the oscilloscope functions. You are correct, though… is better not to. Thing is, there is room and the new scopes are built like they have a computer motherboard built in.. granted I have to read more into… I am guessing lower noise systems.

          2. In regards to the power supply, I have to verify what the source requirements will be from the schematics input requirements and I am not sure buck converters will be clean signal enough.

            I compiled so much information, I was overwhelmed at the time. I still need to read through. Here is a link I made: https://www.youtube.com/watch?v=jZjRVAb3Ykc&t=1s

            I’m more confident higher quality rectifiers for input 24V to whatever down will be easiest. Using a 12V will be more problematic at least off the top of my head if I am thinking correctly.

          3. That and not having to use an inverter to power for off the grid reasons got me thinking I can try not only to repair the power supply if the issue… go ahead and plan on researching a hack for DC operation. Waiting on a determination for a full or partial or no refund for the Realistic PRO-2006 I ordered with memory and squelch mods that doesn’t work other than starting scans from 25Mhz or 1300Mhz and with a large current draw coming from the CPU-Logic Board (buttons not working too so determined bad… though may remove mods and see if changes). Then, I can start the batch soldering work jobs I have lined up (remove 2006 mods, MFJ-203 repairs, 5V&12V dual power supply for DDS Freq/Signal Gen, 520 repairs, semi-rigid cable fab, HackRF RF Shield, solar panels and I think that is all for now).

      2. There is a little interest in running one of the newer Orange Pi or Pi systems, though I am not sure about there capabilities of running Windows and I am not so well versed in Linux systems yet. I am improving my understanding though not a capable programmer yet with Grub/Bash/the different Kernels and files. Soon, though not yet. Still deciding which distros and builds to use. The Ubuntu 17.04 and Kali 2017.1 are what I have used and may stick with, however I don’t want to spend the time custom programming if I don’t have to. I want to test Pentoo out also.

        1. Not really. How many people who read HaD / watch Ben’s channel can afford, or have the budget at work for, a scope like that?

          If they do, they already know the market pretty well.

        2. “That marketing department knows what it’s doing.” They do!

          What they’re doing is putting out their absolute top-of-the-line, drool-worthiest scope visibly in the hands of the people that normal hobbyists look up to, at the same time as they’re offering an entry-level scope, at a bit of a premium relative to the Rigols or used market. They want you to think “I wish I could afford one of those — hey wait, I can! And it’s not all _that_ expensive.”

          They’re not going to compete with Rigol on price in our segment, so projecting the brand image of “quality” as hard as possible is their best angle.

  2. If I understand well the trick with large and small particles, a fourth colour could be added as well. In fact, it seems we can trade off number of colours for refresh time and display thickness, since separation is function of time (speed and path length).

      1. Wow! That does look great. Maybe this means they might actually consider *selling* one someday. Word from the inside is the folks in charge of ink formulation came right from the Duke Nukem Forever School of Management, and have been futzing with the colors for the last 15 years because it’s not absolutely perfect, UV-stable, good to 10^40 cycles, etc. rather than sell what they’ve got in the markets where it fits while they work on improvements.

  3. Slightly disappointing hack. I was hoping for a more involved examination/recreation of LUT entries rather than “junk most of it and I’m sure it’ll be fine”. Still, he’s found some useful datasheets.

  4. Most of that LUT / waveform stuff is for greyscale. If you stick with 1-bit monochrome you can push if much faster, and you can send draw commands to the eink controller even faster by sending commands before previous ones are finished. SOme controllers can process up to 16 overlapping commands, and use tokens to identify which control sequence completed if you wish to run at max speed. However, you can ignore all that and just pace out draw commands at a speed slightly slower than when the display starts to look ugly. He did say his controllers support a 60Hz refresh rate. The eink kindles I worked with supported 40Hz. Older kindles supported (when pushed faster than their spec allowed for) of about 7.3 Hz, which is (surprisingly) good enough for “real time” video:


    1. I was going to say that he mention it as an inconvenience.
      But it seems that, with good timing, and if the screen is consistent, you can indeed have grayscale!
      IMO, that’s almost better than faster refresh rate :).

      1. The waveform tables are also temperature and batch compensated (because of variable oil viscosity), so you need to measure the thermistor when deciding which LUT to use, and not all “identical” displays use the same LUT due to batch viscosity variations. You might need to calibrate your grayscale yourself if needed…

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s