/       /       /    64×32 RGB LED panels, or how to buy this kind of equipment in China
Support the project on Patreon


64×32 RGB LED panels, or how to buy this kind of equipment in China

$10.74 (без учета доставки)
Go to store
This is already my fourth review of LED panels for building displays. This review wasn’t originally in my plans, and neither was buying these panels. In my previous review, I wrote that because of a seller’s mistake, I ended up with a controller I didn’t need. So I ordered another batch of LED panels for it. Since the display was going to have specific dimensions, I had to order panels with a resolution of 64x32 pixels.
This is a short review, with a little theory and answers to some of the questions I was asked after publishing the previous parts.

As I mentioned last time, a mistake in the listed specifications led me to order an HD-D10 controller, which turned out not to support 1/32 scan mode. The seller eventually gave me a $20 discount toward an HD-D30, which I’d already received and put to use.
Actually, I didn’t put it to use myself. The panels were ordered through me, not for me, but I don’t think that matters for this review.

So the board was sitting around with no use for it, and I couldn’t bring myself to throw it away. I decided to buy panels for it and build another LED display. We chose a configuration of 4x3 P3 panels (3x3 mm pixels), for a total resolution of 256x96 pixels. Each panel has a resolution of 64x32.

Having learned the hard way, we looked for panels with 1/16 scan, which are essentially “halves” of the previous 64x64 panels.
The panels were ordered in two batches of six. Having learned another hard lesson, I wrote to the seller that I’d ordered 12 panels in two parcels, but they all had to come from the same batch! The seller replied, “OK, don’t worry.”
I wasn’t particularly worried, really. Last time, that kind of “trick” cost the seller $20 in compensation.

The seller took about two weeks to ship the panels. That’s the longest shipping delay I’ve had, but they arrived in about 10 days, so the total delivery time was almost the usual three weeks.
They were packed “with gusto.” I didn’t think a parcel with six panels could be this big.


Part of the reason for the large package was that the seller didn’t skimp on packing material. The panels were also packed in pairs, so that’s a solid 5 out of 5.


The first parcel contained six panels and a set of cables for them. They even included seven ribbon cables instead of six, which was a nice surprise.


As last time, I filmed the unboxing. I was in a good mood and was about to open the second parcel without filming it, but changed my mind just in time. As it turned out, the second parcel was missing the data cables. The power cables were there, the panels were there, but no ribbon cables. It may seem like a small thing, but it’s annoying. So don’t be lazy—film your unboxings.


At least all the panels were there, and none had any visible damage.


The power cables are quite standard; this is what they’ve always included with the panels. Though these wires look like they might even have a hint of copper in them :)
The usual practice is to include one data ribbon cable per panel; you don’t need any more than that. Power cables, on the other hand, are always supplied at half the number of panels, rounded up, since each cable can connect two panels.
So if you order 8 panels, you should get 4 power cables and 8 ribbon cables. For 9 panels, you should get 5 power cables and 9 ribbon cables.


From my visual inspection, all the panels look identical. I’d guess they’re even from the same batch, so the seller kept their word.


Since these panels are smaller than the previous ones, I could even fit them all on my desk :)


But as I mentioned above, I had a slightly different display configuration in mind, with an aspect ratio of 3:8 rather than 2:3, as shown above.
The display dimensions, excluding the enclosure, work out to 256x3=768 mm long and 96x3=384 mm high.
As a reminder, calculating panel dimensions is very simple: multiply the number of pixels along the length/width by the pixel size, which is indicated in the name: P3 (3 mm), P4 (4 mm), and so on.


All the panels are basically identical. On a few, the power connector is soldered on a little crooked (the top panel), but otherwise they’re the same.


The wide frame makes the panel look bulky, but a wide frame is actually a plus: it means less chance of deformation.
The build quality is pretty good; everything is neat. The signal direction and the top/bottom of the panels are marked.


There are only slightly fewer chips than on the 64x64 panels: those had 36, and these have 31. At first, I thought there were a lot of them.



1. The panel also has markings indicating that it is:
P3 — pixel pitch
2121 — LED type
63*32 — resolution
16S — 1/16 scan
I couldn't figure out the last marking, D1.0; it's probably the board revision.

2. There are three buffer chips next to the interface connector. The signal passes through them to the next connector.
3. The controllers are the same ICN2012 chips as before, four of them.
4. The remaining chips are marked DP5020B. As far as I understand, these are LED drivers; there are 24 of them.


I'll take a short detour here and try to explain how the panel works and what “scanning” means, at least as I understand it.

The signal enters the panel and is passed on to the next panels through the buffer chips.
Along the way, the signal branches off to the ICN2012 controllers, which control the DP5020B drivers.
The LEDs are arranged in a matrix, with horizontal and vertical lines.

Each driver receives both the signal to turn an LED on and the command that controls its brightness. That's how the LEDs are controlled.



Scanning is one part of this process. If we break the whole operation down into cycles, it looks like this:
Cycle 1 — the upper drivers set the signal pattern for all the lit LEDs in the first row, while the ones on the left switch on power to the first row. The result is a glowing strip.
Cycle 2 — we turn off the first row and do the same thing with the second, but the combination of lit LEDs and their brightness now matches the image in the second row.
Cycle 3 — repeat for the third row.
Cycles 4–16 — repeat for the remaining rows.
Cycle 17 — return to the first row and repeat, taking into account any changes to the image.

As you can see, I've shown a 16-row example. This type of scanning is called 1/16. If the scan mode is 1/6 or 1/32, there will be the corresponding number of steps.
Usually, if a controller supports 1/16 mode, it will also work in 1/8 mode, but it will most likely refuse to work in 1/32 mode. As a result, half the rows simply won't be processed, and the screen will show what I saw while preparing the previous review.


Of course, you might ask: if the scan mode is 1/16, why do the panels under review have 32 rows, not 16? It's simple: sometimes panels are essentially “two-in-one” devices. In my previous modification, I showed that there are two red lines, two green ones, and two blue ones. In other words, you could think of it as two 64x32 panels in one housing. Here, we have two 64x16 panels with a combined resolution of 64x32 pixels.
The program settings include a refresh rate option. This is the rate at which the rows switch. Last time, it was set to 960. Divide 960 by 32 and you get 30, which is the frame rate. If you use a panel with 1/16 scanning at the same refresh rate, you'll get 60 frames per second instead of 30. That's the math for you :)


I think we've covered scanning. Let's move on.
The connectors were clearly soldered by hand. You can see flux residue, but the soldering is quite neat, so it's perfectly acceptable.


As with the 64x64 panels, there's no plastic frame like the one on the P4 panels. The front is simply a board covered in LEDs.



The panels are 16 mm thick, excluding the connectors, and measure 192 mm long by 96 mm wide.


The first test was to check for dead pixels. The controllers have a built-in program for this, activated by pressing a button on the controller. First, the whole panel lights up red, then green, blue, and white, followed by various stripes and so on. It's very convenient, since you can test the panels without connecting them to a computer.
I also found out a few things along the way:
1. The controller draws 720 mA.
2. Based on point 1, I calculated the panel's current draw in different modes: red — 1400 mA, green — 860 mA, blue — 460 mA, white — 2710 mA.
3. The power consumption of one panel is therefore: red — 8 W, green — 4.3 W, blue — 2.3 W, white — 13.55 W.
4. When I tried connecting two panels, my power supply went into current limit mode (5.1 A), and the voltage dropped to 3.8 V. But the controller kept working, and the panels kept shining as if nothing had happened. This shows that the drivers regulate the current, and that the controller is quite resilient to this kind of problem.

All tests were run at 66% brightness (I wrote about this quirk in the previous review).


And of course, how could I leave out the video that was in the previous review? :) While I was at it, I measured how much power the panel draws.
I got around 400–800 mA, or 2–4 W, depending on the brightness of the scene. When displaying the splash screen, it drew just 80 mA, or 0.4 W, and I wouldn't say the brightness was low.
So the whole screen, made up of 12 panels, should draw around 24–50 W when playing video and up to 150–160 W when displaying white. The theoretical power could reach 240 W, since the panel was running at 66% brightness.
Based on all these measurements, I'll try to answer one more question: what about heat?
It's quite difficult to predict how hot devices like this will get, since many factors affect it:
1. Ventilation inside the enclosure
2. The enclosure's dimensions, or more specifically, its thickness.
3. The enclosure material.
4. The ambient temperature: being outside in hot weather is one thing, while being indoors with air conditioning is quite another.
5. Exposure to direct sunlight. The panels are black and can heat up just from that.
6. The type of image being displayed. White text on a black background barely heats the panels, unlike black text on a white background.
7. Whether the power supply is built into the enclosure or located outside it.

All in all, if you want to do it properly, you should plan to dissipate around 50–100 W, with peaks of 150 W (for this screen). In actual use, though, the power will most likely be lower, around 30–50 W. If you keep the panel out of direct sunlight and avoid displaying large areas of white, the screen should work just fine outdoors even in hot weather. But in direct sunlight, I'd recommend either active cooling or a large enclosure with ventilation grilles.


By the way, you can make out the image even on a single panel, at least in close-up shots.


For all the tests, I used the HD-D10 controller, which I covered in some detail in the previous part.


I've also been asked how many panels the controllers can drive. There are a few things to keep in mind here, too.

The panel configuration depends directly on the controller's capabilities, and it's important to distinguish between:
1. The number of panels in a row—or, more precisely, the maximum number of pixels in a row.
2. The number of rows, which depends directly on the number of outputs on the controller board and the scan mode. To simplify, the number of rows equals the number of board outputs, while the resolution is the panels' vertical resolution multiplied by the number of outputs.

Here's a simple example to start with: I connected four panels and selected the resolution of a single panel, i.e. 64x32.




The image looks pretty awful, of course. The photo on the left has the 180-degree rotation enabled (left over from the previous configuration).


Set the panel resolution to 64x128 (4 panels), then enable 90-degree rotation. The vertical and horizontal resolutions will switch automatically.


With the first configuration, the image will be vertical.


With the second, it'll be horizontal, which is what I need here.


With the panels laid out as shown, the image isn't very clear. But line them up in a row, and you can see that it makes a perfectly decent screen.


Now let's do the reverse: take six panels and connect them to a single output, then set the resolution to 384x32.




I deliberately took a photo of the opening splash screen, because the image is really stretched out.


Let's try setting the planned resolution. As I mentioned above, that's 256x96.




Since only one row is connected, only part of the image is visible. And since it’s connected to the first output and rotation is disabled, you can see only the very top of the image.


At the time of this test, though, I only had seven ribbon cables, so this was the only configuration I could put together.


While I was at it, I tried a few things just for fun.
I decided to display a 64x32 image on a screen like this. As you can see, the image looks normal only on the first (far-right) panel; the others show some distortion.




Now I’ll try to answer the question: how many panels can you connect?
The screenshot below shows a number of different options, but the key point is how the maximum horizontal resolution changes. You’ll also notice that, despite all my tinkering, the maximum vertical resolution doesn’t change at all. In a way, this is similar to a multicore processor: the board’s processor can handle four (or two, three, five) outputs independently, so it doesn’t matter how many rows are connected.
But since the panels in each row are connected in series, it all comes down to how much data we can transmit, and at what frequency.

1. The basic mode, which is essentially the controller’s default setting.
2. Increase the transmission frequency by 1.5 times. The horizontal resolution goes from 437 to 661 (the manufacturer specifies 512), while maximum brightness barely changes.
3. Let’s try to improve the image quality by increasing the number of gradations from 4k to 16k. This also increases maximum brightness, but significantly reduces the resolution.
4. Set the number of gradations back to 4k, but select the brightness mode. The main thing is, I’m not entirely sure what this means, but neither brightness nor resolution changes.
5. Prioritize screen brightness. At the same resolution, brightness can now reach 94%; resolution increases slightly, but the refresh rate drops.
6. And one last experiment: prioritize the quality of brightness gradation. The overall brightness drops to 35%, and the resolution to 328 pixels.



The controller can actually operate in other modes, with a higher refresh rate and more gradations, but the number of panels connected horizontally will limit what’s possible. For this example, I deliberately chose settings that kept the values out of the red zone.
Here’s an example of what the settings look like when using four independently connected panels (the first example I showed above).
1. You can easily increase the screen refresh rate to 2880, or 180 k/s, with 1/16 scanning.
2. But attempts to improve some of the other parameters push the “Minimum PE” value into the red zone. I still haven’t figured out what that means. This parameter also occasionally went into the red zone in my previous experiments.

So that’s the answer: the number of panels connected horizontally depends heavily on the controller settings, while the number connected vertically depends on the number of outputs. Some controllers (for example, the Onbon bx-5ql) may have six outputs: two for color panels and four for monochrome panels. They can’t be used together, so in this case we count them as two outputs.


After playing around with it, I left the settings as shown in this screenshot, lowering the data transmission frequency to 20 MHz and increasing the number of gradations to 16k.



Now I’ll try to offer a few tips on protecting yourself when buying panels and controllers from China.
I don’t have a lot of experience, but I’ve learned a few things about how to communicate with sellers, as well as the kinds of problems I’ve run into the hard way.

I’ll skip the usual cases—an item never arrives or the wrong item arrives—and focus on the main problems and mistakes.

1. Receiving the package.
Be sure to film yourself opening the package. In one case, I received mismatched panels; in another, some ribbon cables were missing from the package. The cables themselves cost next to nothing, and you can have them crimped at a local electronics market, but it’s a half-hour walk for me, and it might be an hour’s drive for someone else. So film the unboxing—you may at least be able to get compensation for the gas.


2. The panels should all be from the same batch.
Before placing your order, be sure to tell the seller you need panels from the same batch. Otherwise, you may receive panels with different color tints. It can be very difficult to prove anything afterward.
This happened to me when I ordered 12 panels in two separate packages. I saw the same thing with my second order, when I bought some additional panels, but in that case the order was from different sellers, so I didn’t make any complaints.



3. Insist on panels of the same version.
Be sure to specify that you need identical panels. Otherwise, you may receive different ones, as I did. The problem can be resolved, but why go looking for trouble? I mentioned above that something similar happened with another seller, but that one worked out fine. When I ordered the P3 64x64 panels, though, half of them had the colors mixed up. In the end, AliExpress awarded me $20 in compensation; I had asked for $40 for the two panels. The seller got two stars for refusing to admit fault, especially since I had told him twice before ordering that I needed 12 panels.


4. Controllers.
Before placing your order, be sure to ask the seller whether the panels you’ve chosen are compatible with the controller. This won’t guarantee compatibility 100%, but it will cover your back, and if a problem comes up, you can try to get compensation.


In my case, it cost the seller $20 off my next controller. I still gave the seller five stars because he helped resolve the problem and quickly corrected the product specifications on the page, even though the manufacturer still has them listed incorrectly.
As agreed, the seller discounted the second controller and sent it with the same firmware version as the first one.


But there’s no connector for a WiFi card, and the heatsink now sits directly over the chip instead of being offset. The latter confirms the idea mentioned in the comments: the heatsink was offset specifically to make room for the WiFi card. Since that wasn’t critical for me, I let it go.


5. Dead pixels.
Be sure to check the panels for dead pixels. That’s a defect, and the seller should compensate you for it. In the first case, the defect showed up quickly, and the seller offered two options: send a replacement panel with my next order or refund 50% of the cost right away. I chose the second option.


Sometimes a pixel behaves rather strangely, as if it’s saying, “I’ll work when I feel like it, and I won’t when I don’t.”
Video from the review of the first panels.


6. Dead pixels, part 2.
Even if everything looks fine at first and you’re ready to confirm receipt of the order, stop and don’t rush.
Leave the panels on for a while with a white screen at 60-80% brightness; more dead pixels may show up.
Below is a photo of a panel from my latest order. The problem appeared after the panel had been running with a white screen for 10 minutes. The panels had worked perfectly under normal conditions for almost a whole day.
I’ve messaged the seller; he’s thinking it over.


If you’re not put off and still decide to buy panels and controllers from China, here are a couple more tips:
1. Try to buy the panels and controller from the same seller. That makes it easier to sort out any compatibility issues between them.
2. Always contact the seller before ordering. Message them from the page of the product you want to buy so your conversation gets linked to the order and the Ali moderators can take it into account if there’s a problem.

So far, I’ve placed only five (six) orders, and here’s what I can say about these stores.
1. Store from the first review. I bought panels and a controller. The seller was completely reasonable and resolved the problem quickly. Five stars.
2. Store from the second review. I bought panels. I can’t really say much: they shipped quickly, everything arrived, everything works, and I didn’t contact the seller. Five stars.
3. Store from the third review. I bought controllers. The seller was completely reasonable, quickly figured out what the problem was and who was at fault, and was willing to communicate. Five stars.
4. Store from the fourth review. I bought panels. The seller is a real piece of work: first it’s “I don’t understand what you’re saying,” then “It wasn’t me, and it wasn’t my horse,” and finally “Not my problem; I know nothing about it.” In the end, I had to open a dispute and get the Ali moderators involved. I strongly recommend avoiding this seller. If you do buy from them, be sure to put all your requirements in writing before ordering. Two stars.
5. Store from this review. I can’t say anything definite about the seller yet. They’re reluctant to reply, but they do reply. We’ll wait and see what happens with the missing ribbon cables and the dead pixel. The problem may not be the seller’s fault, but it’ll be interesting to see how quickly we can reach a solution.

I’ve just noticed that points 2 and 5 refer to the same store, so there’s a chance to rate both at once.

That’s about it. I hope this review was useful to anyone thinking of buying LED panels and controllers, and helps you avoid the pitfalls my friend and I ran into while placing our orders.
$10.74 (без учета доставки)
Go to store
This page was found when searching for:
панель 64 на 32, характеристики матриц горизонтальные и вертикальные rgb, распиновка 32х64, матрица 32х64 распиновка, китайская led панель обзор, двухстрочные табло ndt rgb 64x32, как называется панель двухслойную в которую втыкают доиоды для свечения, картинка табло 32 на 64, проекты на rgb панелях, Табло РМ светодиодное, семицветное, матричное, Ethernet, RJ45, питание PPоE, 4 символа, высота символа 100 мм, планка для аппликации, режим "бегущая строка", QuickQ WPD M100 9X28RGB, 2, rgb светодиодная матрица 5 х 3, 32X64RGB - Табло РМ светодиодное, семицветное, матричное, Ethernet, RJ45, питание PPоE, 2 строки, высота символа 30/80 мм, режим ""бегущая строка""", https://www.kirich.blog/obzory/drugie-obzory/474-svetodiodnye-rgb-paneli-64h32-ili-kak-pokupat-v-kitae-podobnoe-oborudovanie, схема rgb панели, светодиодная панель 32х32, управление rgb led 64х32 матрицами, rgb 5020, как происходит охлаждение светодиодного экрана, rgb led матрица 64х32 модель ночного светильника со световыми эффектами, pixel strip rgb 64 led, светодиодный дисплей с мини usb для помещений, 390х75х30мм, ip20, синий, ml 961 128 b mld обзор, led панель rgb, информер на матрице 64х32, табло светодиодное 32*64 , rg

Questions and answers

What are the resolution and scan mode of the P3 LED panels in the review?

Each panel has a resolution of 64×32 pixels and 1/16 scanning. In a 4×3 configuration, the screen has 256×96 pixels in total.

What cables should come with the RGB panels?

The usual set includes one data ribbon per panel and one power cable for every two panels, rounding the power-cable count up. The author's second parcel had no data ribbons.

What does the author recommend checking before buying LED panels and a controller?

Discuss panel-controller compatibility with the seller before ordering and send the requirements from the product page. The author also recommends recording the unboxing.

Calculate manuallyAll calculators

Discounted products


You may be interested in


Discounted products


Support the project

All the content on the site and the advice are free, but we’d be grateful for your support of the project and the channel!

Comments2

Comments are published in their original language.

  1. Алекс
    4 December 2019 20:38#1
    Схема у вас неправильная. 2012 управляют выбором строк(какие две строки светятся) - сигналы A,B,C,D.. А 5020 - сдвиговые регистры буферы светодиодов строк- сигналы RGBR1G1B1. 24 регистра 5020. Панель двойная значит на половинку 12 регистров. Три цвета значит 4 регистра на цвет в строке. Регистры 5020 16ти разрядные итого 16*4=64 светодиода в строке по три цвета. Или три канала по четыре регистра на строку половинки дисплея. Идет постоянный перебор 16ти строк в каждой половинке сигналами ABCD и в промежутках заливка буферов двух строк в разных половинках по сигналам RGBR1G1B1. 
    +3
    1. 7 December 2019 04:20#2
      К сожалению я не специалист в этой теме, потому скорее предположил как все должно работать.
      Спасибо за поправку :)
      0

Discuss on the forumHave a question or some experience to share? Discuss the article with the author and readers: the forum is a convenient place to share schematics and photos.
Go to the forum

Only registered readers can comment. It takes a minute to sign up, and you can log in with the same account you use on the forum.

Log inRegister