The source codes are posted #236
Comments
|
Eh used quartz lamps at one time and it was a mistake. They have too much inertia. |
|
"Quartz lamp" is a term very generic, what differentiate IR lamps is their wavelength (Near Infrared, Medium IR, Far IR), so you might have had a bad experience with the wrong lamp for the application. Far infrared lamps (eg, transparent quartz tube with coiled carbon filament heater) are better fit for this application than the current tubes used on the T962A. Far infrared lamps are less color selective and have very short delay. Ceramic IR emitters are better used in conveyor ovens together with convection fans. Check the book "Reflow Soldering Apparatus and Heat Transfer Processes, Elsevier 2020" for more information. |
|
sevstels, Good job!! |
|
The fact is that in industrial furnaces there are many melting zones and we have only one... That's the big difference... Your board falls into a heated chamber and does not heat up from scratch in one chamber... |
|
I agree, if we could afford (both financially and physical space) a multi zone conveyor reflow oven we would probably not be trying to improve a T962 or derivatives. In the early 2000's I modified a toaster oven with two IR top elements and two IR bottom elements. PID controlled using ATmega16pc @ 4 MHz. Serial communication (rs232) with PC and 4 lines LCD display. PWM output was synchronized to the zero crossing of the power line AC and a PWM cycle consisted in a 512 power line semi-cycles time, the pwm power conmutation was equally distributed across the whole 512 (60hz semi-cycles)-cycle using a modified bresenham algorithm, trying to compensate against the IR lamp's inertia and delay. That oven is still in use today. It simply works but the limitation in pcb size made me investigate other desktop alternatives, it is not a multi-zone reflow oven but emulates one "in a good enough way" is is reliable and repeatable (we only use Sn/Pb soldering pastes), so we only change the profile when we buy soldering paste from a different brand. I think that sevstels' work is a step forward into another alternative for all of us who don't have the space available to own the bigger multi-zone reflow ovens. I also appreciate the work involved into developing his advanced software. Don't misunderstand, I am very glad Unified Engineering developed the new firmware for the original board, but we should not be limited by the original board's limitations (eeprom size, display choice, analog thermocouple amplifiers, etc) only two custom profiles' memory, profile customization mode, etc) Some of the limitations are very easily overcame, eg, we could replace the eeprom chip with a bigger memory size Fram chip, same footprint, pin by pin compatible. eeproms have limited amount of write cycles and not that fit for holding user profiles that might change often. Also, the display used is not commonly available nowadays. It also seems that nobody has noticed, or everyone has underestimated, the lack of proper thermocouple connectors in the original and on the subsequent "improvements",. Thermocouple connectors are not made only to inflate the bill of materials, they ensure that there are no other thermocouple-junctions in series besides the "cold junction" everybody now cares about. By using unknown metal contacts in the connection to the board we are introducing a big source of noise and error in the measurements by creating extra junctions. |
|
mikhailmihalkov, I saw photos of your mods, on another thread, using ruby IR lamps (near infrared), how was your experience? |
Of all types of lamps, these are the fastest.... They hold a thermal profile very accurately and have the smallest inertia they are perfectly friends with the convention.... Unlike those supplied from the plant.... I have already redone three furnaces 962 ,967.... |
|
Thank you, Congratulations!! |
|
In my opinion, the speed of quartz lamps is enough. It is important to understand the basic principle - there should be a lot of lamps... So that the array of lamps works without significant overheating relative to the board temperature and PID regulator does not enter into saturation. It is important that the coil heating is as low as possible and the lamps emission is as long wavelength as possible and does not go into the visible red or orange range of the spectrum. If you have only one or two tubes in your furnace - to heat up the contents of the soldering chamber, you definitely have to overheat them considerably! In this case, the performance of the system will be the worst. Almost all single chamber furnace modifications that I have seen contain this very error. The lamps are few and glow in the visible part of the spectrum. In a "proper" soldering furnace, you shouldn't be able to see the operation of the lamps with your eye. And so, if you have installed a lot of additional lamps, the dependence on their speed will no longer be critical. The power will be enough to heat up the board with long wavelength radiation only. The main delay now will be the thermal inertia of a board with massive components. Such a board heats up slowly and cools down slowly. If you try to give more power to strictly follow the profile the lamps will go from the long wavelength range to visible red and the shorter wavelength light will melt and ruin the plastic of the component housings. This should not be allowed to happen. Therefore, when heating, you should move in small increments of temperature increase, e.g. +2 degrees. Take another step and wait until the temperature is established. The temperature step should be dynamically recalculated, automatically dividing the interval remaining to the point of extremum by the desired amount of temperature increment. It is possible to install an additional sensor for the intensity of visible radiation and limit the heating by it. In this case there will be no overheating of the components by the short-wave light radiation. Because the lamps will be in a small incremental current mode relative to the previous step and the board will warm up qualitatively and evenly. The temperature profile will strictly follow the temperature, but the time of its movement will dynamically change. |
|
sevstels
|
|
I threw out the original heaters. They were of very poor quality. Their spirals were not wound evenly. There was some sand inside the lamps, so the inertia was very high. You should use normal heaters, with a lightweight quartz tube and a thin coil. My new heaters were 300 watts with a thin coil. That's why they are fast enough. For cooling should replace the fan with 4 times more powerful. But for the most part it doesn't matter. |
|
If you have the desire and skills to help us make the T962 app for Android. Please write me to coordinate your efforts. |
|
I just received my T962A, opened the top cover and got rid of the visible masking tape. Since my oven is a little bigger than the T962, I can't find the proper lamps to try sevstels' IR tube arrangement yet (I also need to do the proper spacing calculations for my oven), but first I need to keep on disassembling and make some measurements including the lamp sizes, distance to the pcb placement location on the drawer, etc As I was commenting previously, I did a toaster oven modification, for reflow purposes, about 20 years ago. I still have it, never bothered to replace the right lateral covers but it works fine for my needs. I remember that one of my concerns at the time was trying to heat the pcb form the bottom too, so that the temperature gradient between top and bottom was limited to only about 30-40 degrees (worst case), therefore I also included the bottom heaters into independent PID control. Thermocouples on top and bottom of the pcb. Bottom IR tubes were PID controlled up to just the final of the soaking stage, from there up the top tubes' PID (and top thermocouple) took control of the oven's heating until the final of the profile. I can't find now literature to backup the need to limit the temperature gradient between top and bottom of the pcb, but having bottom heating also helps a lot when the top tubes are limited in power. The pcb to be reflowed rests on top of a wire grid instead of a tray, so it is easy to get IR heat evenly. PCB sizes are limited to about 10 x20 cm, the maximum (rated) power is 1500W. Not enough for unleaded solder paste profiles because I also never bothered to insulate the small oven chamber. No convection fan involved, cooling by opening the door slightly. Profiles were saved in a monitoring pc, also each board's reflow logged data was saved in order to help investigate possible failures if there were problems when deployed. |
I agree. Bottom heating is necessary. |
|
quote: "..I can't find now literature to backup the need to limit the temperature gradient between top and bottom of the pcb" ... |
|
|
Natural convection helps when there is a lower heating element to heat the pcb from the bottom, unfortunately using only top heating elements (without forced convection) colder air tends to accumulate below the pcb level so contributing to create the vertical temperature gradient that produces pcb warpage. The bigger and thicker the pcb the most probability of warpage in presence of temperature gradients, if the oven has horizontal hot spots, within the pcb area, those will also contribute to local warpage. The pcb substrate material also plays a role on the propensity to warpage. |
I think we all are talking about the same subject in different words. Yes, the above statement is correct for all the IR lamps to produce far IR radiation depending on the filament temperature, the problem is efficiency, predominantly near and medium IR lamps will require a power limit in order to produce more IR in the far wavelength range (by dimming the filament glow because of a lower filament temperature) but, in order to get the same thermal effect (irradiation), you will need to add more lamps, If you use the same power lamp with predominantly long wavelength (ceramic radiators or carbon filament quartz tubes) you will have more radiation in the far IR zone with the same power without adding more lamps and without having to limit the power. Ceramic radiators have too much inertia and are slow responding so I don't consider them candidates for this oven. I also agree with sevstels calculation to get an homogeneous radiation coverage by placing the heaters at the right distance between lamps, and in order to cover more area it will be necessary to add more heaters. He had to limit the power in order to get most of the radiation in the far infrared range using standard quartz tubes |
|
I do double sided boards and the oven after the mods works well.
if there was heating on the bottom doing double sided boards without
epoxying the bottom components would not work
…On 11/23/2022 4:40 PM, WalterMurphy wrote:
Natural convection helps when there is a lower heating element to heat
the pcb from the bottom, unfortunately using only top heating elements
(without forced convection) colder air tends to accumulate below the
pcb level so contributing to create the vertical temperature gradient
that produces pcb warpage. The bigger and thicker the pcb the most
probability of warpage in presence of temperature gradients, if the
oven has horizontal hot spots at the pcb level those will also
contribute to local warpage. The pcb substrate material also plays a
role on the propensity to warpage.
—
Reply to this email directly, view it on GitHub
<#236 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAWNV3JYTFDP3E5WGD5JS6TWJ2TQVANCNFSM6AAAAAASAYJMIQ>.
You are receiving this because you are subscribed to this
thread.Message ID:
***@***.***>
|
|
I also do double side boards using a 2top/2bottom IR elements toaster oven (no fan), one side at the time, first the least populated side with the lighter components (normally the bottom side of my pcbs), on the second round I do the other side. The already reflowed components now on the bottom side of the pcb (resting on the support grid) won't fall down. They are normally capacitors and some resistors. If you had heavier components on the lower side I suggest to use epoxy. I solder the through hole big capacitors, connectors, transformers, etc manually. My oven use separated PID control for the top and bottom heaters. The bottom side won't get heated to the peak reflow temperature, those bottom heaters will be turned off after the end of the soak phase, about 20-30 seconds before reaching the peak temperature on the top side. I use leaded paste, that could be a factor. |
|
Don't come up with a bike when it was invented) Look at the designs of branded proven furnaces with one zone |
I did, most of the so called "batch reflow ovens" are forced convection type, there are some made in France with very bad reviews in EEVblog due to safety issues and poor engineering solution from the manufacturers. Found some made in Japan, which have a motorized tray, they use far infrared heaters with no convection, also are ready for nitrogen atmosphere during reflow. The problem is that there are no other specs or disassembly diagrams in their website. They also use an enclosure around the oven itself, probably to contain fumes. I was thinking that applying sevstels calculations instead of adding more tubes, with the right spacing, I can also keep the current amount of tubes, the current spacing, and increase the distance from the tubes to the tray. I know that, following the inverse square law, the W/cm2 at the pcb level will diminish, and I will also require to change the current four tubes for more powerful ones to compensate. I need to crunch some numbers to see which option makes more sense economically. Maybe, since I am open to modify the whole enclosure, I can also add a bottom chamber with convection heaters and fans to do a conversion to hybrid IR/ convection oven. |
|
In general, tested furnaces are those that work in real industries in laboratories, etc... If you do not have the opportunity to look yourself, contact the forums where the contract production works.... This applies not only to furnaces, but also to smd installers.... Such stoves as in this article are the bottom of the market (payment for a low price) If we talk about the most affordable option that actually works and is available to anyone, then this is an example https://www.instructables.com/DIY-REFLOW-OVEN/ the heaters of course change |
|
The heaters on this stove are also wrong. |
|
6 halogen lamps included in 3 sequentially solves this problem |
|
It was quite difficult to buy the GBU521 dongle. It has already been discontinued. I was able to connect a bluetooth dongle on a TI CC2540 to control the oven from a PC. There are plenty of them on aliexpress and cheap: https://www.aliexpress.com/item/1005001999985901.html |
Can you be more specific ? details, Code? |
|
If you are interested to see it as it is, you can download the project from here: https://github.com/sevstels/T962-PC-CC2540 To connect to a Nordic UART, you need set the the BT device address Use a TI cc-debugger programmer: https://www.aliexpress.com/af/cc-debugger.html Example can see here: https://t1.daumcdn.net/cfile/tistory/2161273557831B1D2B If the Nordic nRF51822 address is correct, you will connect in 1-3 seconds after the program starts.
|
If you want, you can try to figure it out and modify it to suit your needs.
Project -> description page -> App
The text was updated successfully, but these errors were encountered: