Author Topic: Hakko still the best option for a good quality hobbyist soldering station?  (Read 4145 times)

0 Members and 1 Guest are viewing this topic.

Online KL27x

  • Super Contributor
  • ***
  • Posts: 3614
  • Country: us
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #75 on: September 18, 2019, 09:04:30 pm »
Quote
In that video he says "keeping the tip absolutely at the right set temperature". Yet the display is magically sitting on 200C the whole time and no tip measurement was actually taken to prove this.
Did he? LOL. Right you are, on all points.

I stopped listening to him by that point and was just watching the scope. 

Quote
If they've used a relatively large thermal mass crimp to create the hot end junction of this thermocouple, that could allow its temperature to be dominated by the heat from the tip itself
Theory sounds ok to me. Except yeah, the crimp/heatsink would ideally be large in surface area and high thermal conduction/cross-section, but it should have low thermal mass as possible. Temp gradient will build at electrical insulation between this crimp/heatsink and the rest of the tip in accordance with this mass, and you want the dog wagging the tail, not the tail trying to wag the dog. So the idea is chasing its own tail, a little? (too far? :) )Dunno what material does that, and then there's still the mass of the end of that thermocouple and heat conducted to/through it from the heater.
« Last Edit: September 18, 2019, 10:49:56 pm by KL27x »
 

Offline Johnny B Good

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: gb
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #76 on: September 21, 2019, 10:54:59 pm »
Quote
In that video he says "keeping the tip absolutely at the right set temperature". Yet the display is magically sitting on 200C the whole time and no tip measurement was actually taken to prove this.
Did he? LOL. Right you are, on all points.

I stopped listening to him by that point and was just watching the scope. 

Quote
If they've used a relatively large thermal mass crimp to create the hot end junction of this thermocouple, that could allow its temperature to be dominated by the heat from the tip itself
Theory sounds ok to me. Except yeah, the crimp/heatsink would ideally be large in surface area and high thermal conduction/cross-section, but it should have low thermal mass as possible. Temp gradient will build at electrical insulation between this crimp/heatsink and the rest of the tip in accordance with this mass, and you want the dog wagging the tail, not the tail trying to wag the dog. So the idea is chasing its own tail, a little? (too far? :) )Dunno what material does that, and then there's still the mass of the end of that thermocouple and heat conducted to/through it from the heater.

 It wasn't a scientifically rigorous investigation was it? The scope evidence, whilst not so elegant a display as you'd see on a modern DSO with all of its fancy triggering features, did reveal the varying duty cycle of the controller in response to the heatsinking effect of the damp sponge just the same.

 As for the "Tail wagging the dog" expression, it wasn't 'too far' at all. In this case, it's not so much "A dog with two tails" as more a case of "A tail with two dogs". As long as our measuring point is closer to the dog of interest (the tip) than the other, less interesting dog (the heating element), we're in with a chance of measuring what the 'interesting dog' is doing to a reasonable level of accuracy.

 I was searching for any articles referencing the clone T12 tips versus the Hakko originals topic when I came across that dangerous prototypes forum again here, which, if you haven't already done so, I think you may be interested in reading :

http://dangerousprototypes.com/forum/index.php?topic=5264.0

 The use of a 24vac supply to the tip heater does have the merit of allowing continuous monitoring of the thermocouple voltage at the rate of 100 or 120 samples per second and, perhaps more importantly, the minimum of galvanic corrosion due to residual moisture and interference to the thermocouple voltage from any such stray galvanically generated potentials (assuming the power is PWMed in whole cycles of the ac supply).

 DC power tends to create a runaway effect on any such aqueous chemical reactions until, with sufficient heating, the moisture is eventually driven out to halt or reverse this process. It does occur to me that this 'burning in' requirement may simply be a reflection of a galvanic effect from residual moisture which may be at higher levels in the clones and fakes than in the more carefully fabricated Hakko tips.

 Otoh, it could be an effect only apparent in these T12 soldering stations such as the KSGER unit and its various copycats with their use of a 24vdc drive in place of the 24vac used in the Hakko soldering stations and its clones with the effect not being exclusive to the fake and clone tips. Using genuine Hakko tips might get you better quality but it won't necessarily be free of this 'burning in' effect seen with these DC powered T12 soldering stations. It's just a thought and it could all be just more bollix from yours truly. :-//

 The plain fact is, I haven't been able to unearth any definitive tests between fake/clone tips and genuine Hakko T12/T15 tips in regard of this 'burning in' phenomena observed with the KSGER T12 controllers so I'm left to hypothesise about possible/plausible mechanisms (the Devil makes work for idle hands - in this case, my need to better understand what's actually going on).

 I can certainly imagine the clone tip manufacturers leaving out the final drying out phase that Hakko take care to include in their own manufacturing of these tips. It may simply be a matter of leaving this final finishing phase as an end user instigated 'burn in' phase - just a minor inconvenience for the end user to deal with as far as the clone tip manufacturers are concerned (after all, it's just the 'price' the cash strapped hobbyist end user will be prepared to pay for such 'cheapness').

 In view of the ageing phenomena of thermocouples in general, the tip calibration feature is an 'essential' in all of these soldering stations simply as a means of overcoming the reducing sensitivity with age of the built in thermocouple in these direct drive cartridge tips regardless of brand.

 The KSGER oled display T12 stations have a more complex calibration process than the cheaper 3 digit led display versions, involving as it does, test temperatures of 450, 350 and 250 deg C. In view of the linearity of K type thermocouples and its close cousins over this range of soldering temperatures, this seems unnecessarily complicated a procedure. One can only hope that this extra complexity confers an actual benefit beyond allowing the controller to average out operator measurement errors.

 Although I haven't had as much experience using the KSGER soldering station as I'd have liked by now, it does seem to offer more for hobbyist use than a genuine Hakko FX-951 based setup. At least you still have the option to use genuine Hakko T12/T15 soldering tips if you feel the clone tips are a letdown in some way.

 Mind you, I think it's still worth investing a little less than the cost of a single Hakko tip in a 'starter pack' of ten assorted clone tips if only to get a better idea as to which two or three hakko tip types would best suit your needs, starting with just a single Hakko tip to verify that you'll actually experience a tangible benefit before investing another 50 or 60 quid in another couple of  Hakko tips (they're bloody expensive here in the UK!).

JBG
« Last Edit: September 21, 2019, 10:56:30 pm by Johnny B Good »
 

Online KL27x

  • Super Contributor
  • ***
  • Posts: 3614
  • Country: us
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #77 on: September 21, 2019, 11:22:28 pm »
Quote
It wasn't a scientifically rigorous investigation was it? The scope evidence, whilst not so elegant a display as you'd see on a modern DSO with all of its fancy triggering features, did reveal the varying duty cycle of the controller in response to the heatsinking effect of the damp sponge just the same.
Yeah, it seems to vary pretty sharply. But for all I know it's cutting back sharply to a very low duty cycle at X. And then under no load, the temp will slowly drift up 10-20C higher than X to a "soft" set point.* Then under a load it will sag to some varying degree in relation to the load.

Quote
As for the "Tail wagging the dog" expression, it wasn't 'too far' at all. In this case, it's not so much "A dog with two tails" as more a case of "A tail with two dogs". As long as our measuring point is closer to the dog of interest (the tip) than the other, less interesting dog (the heating element), we're in with a chance of measuring what the 'interesting dog' is doing to a reasonable level of accuracy.
In this case, you'd be making a trade. To increase the thermal coupling to the tip, you're increasing the mass of the sensor. So now you're creating lag. It will be have less error so long as you give it time to stabilize, but it has to measure a moving target. So sometimes the error will be less, but sometimes it will be greater. And sometimes it might be error in the opposite direction, even.

The ideal sensor would approach zero mass, so that even though it is coupled to the tip through an imperfect conductor, even a slow amount of heat transfer will quickly bring the sensor closer to the temp of the tip. We don't need to overcome the "VFD" of the conductor/insulator that couples it. The little trickle of heat down to something approaching zero differential will keep it close enough for the job. Big dog easily wags small tail. Dog doesn't wag the 50 lb tail very effectively. For a thermocouple, for instance, the ideal shape would be a thin foil with a large surface area to maximize coupling to mass ratio. Perhaps a thin layer of metal vacuum deposited onto an insulator to plug up the hole in the middle with what acts like near zero additional thermal inertia once the insulator in the center finally gets to operating temp. (Yeah, I'm just making shit up at this point).

I said before that I don't know that it's impossible to correct this problem through software. It's just I can't figure out how it is possible. Average person might just assume someone smarter figure it out. I don't roll like that. If I can't figure out how it's done, I tend to believe it only when I see it. If you don't understand the stuff I wrote, it might be my fault.  I am not a physicist; and I'm a bad teacher; I assume you know some of this to understand what I'm saying. (And if you don't understand it, to you it's just one person's "belief system" vs another.) But I understand it; at least I think I do, and the people that come up with standardized aptitude testing also would guess I'm pretty good at this if their test works. What I have described was not some theory strung together to support a narrative. I have an understanding of the entire system start to finish (maybe I made some errors or have left some significant gaps; but AFAIK, my understanding is fairly complete and should be pretty accurate). Plus c/mon. The kicker is I tested it, and the results happen to support what I am describing. I agree though. If someone knows something but can't explain it to me, then I don't fully believe it until I see it. It's fine.

Now, even I don't trust my own bias. I made a conscious decision to use (one of) my T12 irons for a significant period, first. Ended up being 2 months, using is as my primary iron, to get some fuller/fairer sense of the perhaps not-so-obvious advantages and to see if any disadvantage would be noticeable/significant in practical use. It was two months later when I finally did the sag measurement/test. Maybe this sag is not as big a problem as I expect it might be, and if I measure/observe it, first, then I will be biased in evaluating if it makes a difference in performance. So I actually observed the performance con first, then verified it with testing.

One of the simplest tests with no room for subjectivity is the one I did in the thread 2 years ago. Compare two irons with a temp tester on the same heatsink. Adjust until they hold the same temp in this equilibrium. Take the iron away and let it stabilize under no load, and then measure the set temp under no load.

*I'm not suggesting this behavior, randomly. This is the behavior I directly attribute to heater/sensor coupling. This is what I expect to occur. If strictly working with current data with no memory, the circuit could not accurately tell the difference between one and the other. It doesn't know it is slowly drifting up to a no load set point. It doesn't know it's sagging under load. It doesn't know it cuts out too early when heating back up. To the circuit, it would all be "at set temp; all good; did my end of the job as best I can with the given information."
« Last Edit: September 22, 2019, 12:42:45 am by KL27x »
 

Online KL27x

  • Super Contributor
  • ***
  • Posts: 3614
  • Country: us
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #78 on: September 22, 2019, 01:43:00 am »
Resolution.
Ok I forgot one thing. And that might have been due to personal bias. Resolution of the signal.

I previously stated:

Quote
If strictly working with current data with no memory, the circuit could not accurately tell the difference between one and the other. It doesn't know it is slowly drifting up to a no load set point. It doesn't know it's sagging under load. It doesn't know it cuts out too early when heating back up. To the circuit, it would all be "at set temp; all good; did my end of the job as best I can with the given information."

But to be fair, in conjunction with a second sample after a brief off time, you have essentially all the necessary data to calculate or cross check the tip temp. It's just you've got a graph SensorOutput over Time (since heater turned off), where the TrueTemp lines converge or nearly converge at Time = 0, and you're looking at a small slice of this graph right at this pinched up converging end. I made the mistake of thinking this was necessarily a problem. To resolve this you just need to... well, you need to be able to resolve it.  |O. It's so obvious in hindsight. If you have enough resolution, then what is the problem? In practice, you might end up with some unpleasant noise, but even if you did, you could probably average it out and still be fine.

So given enough resolution, and some minimal programmed algorithm involving some calculation and/or some simple lookup tables, I think you could overcome this. Is this the reason the Pace ADS uses a 17-bit ADC? When the heater shuts off, the temp drop would be relatively fast, so you would need a pretty substantial dynamic range to cover that slope. Add what you need to cover the entire temp range while heat is on and while off. And you'd have to still be able to resolve very small differences in the end points of your sensor temps, and even finer differences in the starting points. Seems like it would require a lot of resolution. And it would sure make sense for the display to simply read "low" until it gets at least up to 150C, so as to not waste any of it.

You might still also want to know how long the heater had been on for. That initial first X duration after switching heater state (near max duty cycle for that tip or near zero would be the only two states you would need) would have some differences. If this X duration significantly exceeds the sensor read period, then you would ideally want to account for it. With this plus the "high enough" resolution, you probably should be able to do it, but I'd have to chew on that for a bit to recognize the next thing I overlooked.
 
« Last Edit: September 22, 2019, 09:08:12 am by KL27x »
 
The following users thanked this post: Johnny B Good

Offline Johnny B Good

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: gb
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #79 on: September 24, 2019, 05:28:28 pm »
Resolution.
Ok I forgot one thing. And that might have been due to personal bias. Resolution of the signal.

I previously stated:

Quote
If strictly working with current data with no memory, the circuit could not accurately tell the difference between one and the other. It doesn't know it is slowly drifting up to a no load set point. It doesn't know it's sagging under load. It doesn't know it cuts out too early when heating back up. To the circuit, it would all be "at set temp; all good; did my end of the job as best I can with the given information."

But to be fair, in conjunction with a second sample after a brief off time, you have essentially all the necessary data to calculate or cross check the tip temp. It's just you've got a graph SensorOutput over Time (since heater turned off), where the TrueTemp lines converge or nearly converge at Time = 0, and you're looking at a small slice of this graph right at this pinched up converging end. I made the mistake of thinking this was necessarily a problem. To resolve this you just need to... well, you need to be able to resolve it.  |O. It's so obvious in hindsight. If you have enough resolution, then what is the problem? In practice, you might end up with some unpleasant noise, but even if you did, you could probably average it out and still be fine.

So given enough resolution, and some minimal programmed algorithm involving some calculation and/or some simple lookup tables, I think you could overcome this. Is this the reason the Pace ADS uses a 17-bit ADC? When the heater shuts off, the temp drop would be relatively fast, so you would need a pretty substantial dynamic range to cover that slope. Add what you need to cover the entire temp range while heat is on and while off. And you'd have to still be able to resolve very small differences in the end points of your sensor temps, and even finer differences in the starting points. Seems like it would require a lot of resolution. And it would sure make sense for the display to simply read "low" until it gets at least up to 150C, so as to not waste any of it.

You might still also want to know how long the heater had been on for. That initial first X duration after switching heater state (near max duty cycle for that tip or near zero would be the only two states you would need) would have some differences. If this X duration significantly exceeds the sensor read period, then you would ideally want to account for it. With this plus the "high enough" resolution, you probably should be able to do it, but I'd have to chew on that for a bit to recognize the next thing I overlooked.

 You do realise that in essence, you've just described a PID (Proportional-Integral-Derivative) controller algorithm.  :) see the wiki article here https://en.wikipedia.org/wiki/PID_controller

 I've always assumed that such control can overcome the imperfections in monitoring the tip temperature, given enough sophistication from a full understanding by the programmer of the many factors involved in such monitoring and control. The only 'unknown' in this case being the error between the thermocouple and the actual tip temperature due to manufacturing tolerances, making the need for a calibration routine of vital importance to obtain reasonable thermal control of the soldering process.

 One point I did pick up on from that dangerous prototypes thread, was the observation of the ~20μV per K thermocouple sensitivity used by the Hakko T12 tips (and, of necessity by the fake/clone tips for the sake of compatibility) instead of the 41μV per K of the classic K type thermocouple I had expected to see.

 It looks like Hakko may have chosen something similar to the N type TC simply to minimise the ageing effect typical of a K type. The observed sensitivity is less than the N type which was originally designed as a superior version of the K type so looks to be a proprietary combination of alloys chosen for low temperature coefficient in the wire used to form the heating element and minimal ageing effect. It's probably a lower sensitivity variant of the N type that lends itself well for use as a combined heating element and thermocouple.

 In this case there's no need to create a thermocouple to an industry standard μV per K sensitivity rating, just one that can provide sufficient sensitivity, capable of acting as a heating element. Indeed, it's in Hakko's best interests to use a proprietary thermocouple that best meets their needs with a minimal ageing characteristic. The soldering station can readily be calibrated to match whatever Hakko have chosen by way of the μV per K sensitivity of the built in thermocouple so doesn't represent an additional cost in its production.

 The KSGER's user mediated calibration feature is a rather neat way to address not only variations in clone and fake T12 tips but also that of the original Hakko T12/T15 tips as well as any errors within the controller itself. The default settings might put you in the ballpark with Hakko tips but even here, let alone in the case of the clones and fakes, it's worthwhile calibrating your collection of tips against a reasonably accurate K type thermocouple based tip thermometer such as those Hakko FG-100 clone thermometers which are no less accurate than the original Hakko units costing some 18 to 20 times as much. As long as you're within +/-10 deg C of the indicated temperature which you can fine tune anyway, you're good to go with most any soldering task you're likely to face in a hobbyist or semi-professional context.

 At the end of the day, the experienced solderer will adopt settings that best suit his working style and the nature of the soldering task that confronts him. The numbers on the soldering station merely represent a starting point of reference from which to fine tune for optimum performance anyway. Aside from this need to 'burn in' new tips when using the KSGER station, it seems to me a better alternative to the Hakko FX-951 station as far as the hobbyist/semi-professional user is concerned although I have to admit that I have little working experience to base that opinion on as yet.

 The issue with brand new tips may be nothing more than a need to drive out any residual moisture which can provide ionic conduction for the applied DC to create galvanic interference from the products of electrolysis so formed until all the moisture has been driven off to halt or reverse this undesired effect and it does occur to me that it might be worth rigging up a suitable 12.6vac or 15 vac 'conditioning' supply to drive out such moisture sans the DC bias voltage of the KSGER soldering station heater current.

 Since there is no such thing as a "failed experiment" (just an unproven theory - all experiments will tell you something useful, even if it's not what you'd hoped to discover), I'm looking forward to seeing the results (once I've located a suitable transformer). It might prove a better way to "Burn In" these T12 tips regardless of whether they're genuine or fake. Besides which, if such a heating cycle to drive out any residual moisture does work, it'll speed up the calibration process which, at a starting temperature of 450 deg C, will be no bad thing in terms of the needless tip oxidisation involved when I have no intention of subjecting them to such abuse in the first place.

 I had hoped to make a start in using the soldering station in ernest later today after finally taking delivery of a Mustool G600 LCD display microscope from Banggood earlier this afternoon but, unfortunately, the rear glass panel of its display was broken, either a manufacturing defect or shipping damage despite there being no other obvious signs of such. Thus I am further delayed in putting the soldering station to use and left awaiting a reply to my request for a full refund or a couriered exchange within the next ten days before I order from an ebay seller located in Ireland for a few quid more with a more realistic 3 to 10 day delivery time.

 At my time of life, I'm beginning to realise that I can't really afford the protracted delivery times of Banggood's cheap pricing, especially when what I've ordered, and waited so long for, arrives as 'broken goods'.  >:(

JBG
 

Online KL27x

  • Super Contributor
  • ***
  • Posts: 3614
  • Country: us
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #80 on: September 24, 2019, 07:00:43 pm »
So you consider what I described as PID? Hmm, I guess I didn't think about it that way, because all this is just to figure out the current temperature of the tip. I suppose the temp of the tip is sorta the future temp of the sensor/heater in the moment you shut it off. So yes. But in this sense of the word, all this PID does is to make the iron able to perform like a regular iron, so far. Just to actually read the current temp of the tip. I suppose it makes much more sense now, thanks. I always wondered why people thought soldering irons needed PID, and maybe I was looking at it wrong. Some irons can't be improved with PID, but this one does damn well need it.. to read the tip temp.

So this requires a bit of resolution.

A regular iron need say 300C degrees of range (arbitrary here, lets say 180C to 480C). So it needs a resolution of about 300 in order to keep the iron to within 1 degree. Which is 8 or 9 bits of ADC resolution.
The T12 iron would need much more. How much more? Let's take a stab.

The floor would be the same for either. When heater is off for awhile, it will reach temp of the tip. So say at the bottom of your range around the 180C mark and under the right condition (say the user reduced the temperature setting, so the tip is drifting down to temp), the floor is the same. So let's go up 300 degrees and look at the ceiling.

At 480C, the regular iron, assuming it uses an ADC, needs that range of 300 to get there. The T12 would need that 300C degrees plus w/e the temp of the heater gets to beyond that. So let's add at least 100 degrees C, which is modest. Now we're up to a resolution of 400.

That's not so bad, we got that covered with 9 bits, still. Right? No, we don't.

The iron with the "easy" temp sensor is reading the temp. It needs to resolve down to 1 degree to know what the temp of the tip is to the nearest degree. The T12 iron needs to resolve a range of say 50C true temp tip that happens within a span of... who knows. But that won't stop be from guesstimating a few degrees. Free to add your own guess.

3C divided by 50C is 0.06C; Clarifitation: I'm saying that at the time the sensor is read, within some 100 milliseconds duration, difference of 3C at the sensor at that time would be a 50C difference in actual tip temp.  To know the tip of the temp to within 1 degree, the T12 ADC has to be able to resolve 0.06C changes. So in this guesstimation, it needs about 16.7x as much resolution, nominally. It needs a resolution of 6400, or about 13 bit ADC, at least.

Quote
As long as you're within +/-10 deg C of the indicated temperature which you can fine tune anyway, you're good to go with most any soldering task you're likely to face in a hobbyist or semi-professional context.
Mind you, this is not simply in order to be able to set the temp in 1 degree increments. It's so that the iron can tell the difference between degrees. What it can't resolve, it can't tell the difference. And then the sag comes back into play up to that unit of resolution. If the station is off in calibration, I don't care. If it can only be set to 5C increments, that's fine, no problem. If the unit sags 5C more than the other, then that kinda sucks, IMO.

I think with a 12 bit ADC and very optimized it might be possible to get within 2 degrees, if lucky. But that is if my guesses are close and everything works out conveniently sorting into the boxes we want to stick them in, and if my 100 degree overheat for the heater is enough. 12 bit is still borderline, with my guesswork. Pulled directly out of the rear, obviously. But real measurements could be done and this estimate could be done using real numbers.

I would guess Pace uses 17bit ADC because each of those bits is useful. (Or maybe they got a good deal). So to heck with completely guessing. Short of doing these experiments and measurements, I would err towards "Pace probably figured it out, already." To get your iron to regulate as tight as the Pace, I would bet you need most of those 17 bits. And I don't know any of this. I've never seen an ADS in person. I have only read some other user's observations and posts.

Just guessing in the dark, here. Nothing to see.

« Last Edit: September 24, 2019, 08:32:55 pm by KL27x »
 

Offline Johnny B Good

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: gb
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #81 on: September 25, 2019, 12:20:42 am »
 You may be overthinking this. :)

 I was thinking that this could probably be done with a modern cmos Z80 cpu, say one clocked at a modest 8MHz. 8 bit unsigned integer could cover the range 150 to 400 deg C where you'd assume an overflow (carry result) to simply indicate "Too bloody hot! Cut heater power, buddy!" and a zero would indicate too low a temperature to solder with. You could equate each integer value (0 to 255) to 2 deg C (a one degree C accuracy is way more than we actually need - 5 deg C increments would be accurate enough for this task) which is enough to go from 100 to 610 deg C.  :)

 Even for the original 2MHz clocked Z80, a millisecond time interval is sufficient to carry out over a hundred instructions (provided not too many stack ops are involved) and the rate of change in temperature is unlikely to top half a deg C per millisecond. I'm not claiming to be up to the job of coding a sophisticated PID algorithm in Z80 assembler, just guessing that a basic control algorithm could be cobbled together for even a 2MHz Z80.

 However, with that thought in mind, I began to wonder about the STM32 controller actually used in these KSGER soldering stations so made my way over to the wikipedia article on STM32 controllers here:

 https://en.wikipedia.org/wiki/STM32

 My, oh my, there's a shitload to choose from isn't there?  ::)

 I still wasn't too keen to open up my own KSGER unit so undismayed, I searched with "which stm32 is in a KSGER" and found this fascinating EEVBlog thread, started by floobydust back in June (last entry being 16 July - fresh by my standards :)) :

 https://www.eevblog.com/forum/reviews/t12-stm32-v2-1s-soldering-station-controller-schematic-etc/

 It's only one page long at the moment but looks ready to tip over to a second page. I think you might find it an interesting read. >:D

 Drat that "Pandora's Box" thread! It looks like I will have to open mine up, after all ::)

 I wasn't particularly interested on the OSS aspect, more about the hardware issues, most of which I had already dealt with, including (just for the sake of tidiness) soldering the rotary controller tags to suppress possible ESD issues (assuming the TPHs actually connect to the ground plane - something else for me to check ::)).

 Incidentally, after downloading the reverse engineered schematic, I was rather startled to see the rotary encoder's pulse outputs represented by mechanical switch symbols but, after checking out the Bournes datasheet pdf on these encoders, it turns out that this is exactly how it's done  :wtf: Just two mechanical switches phased to produce overlapping square wave pulses just like you'd see with a simple optical encoder to drive a JK flipflop to synthesise a direction of rotation signal as in a tape footage counter (sans the needless expense (and the accuracy compromising effect) of a rubber belt drive... Akai you blithering idiots! - sorry for the mini :rant: - it's just something about Akai's flagship GX747 tape deck that's been bugging me for more than three decades since I discovered such blithering idiocy of design).

 I noticed a reference to a Hakko patent on this combined thermocouple and heating element tip cartridge direct drive design which, to my great surprise has now expired as of yesterday :wtf:

 Interestingly, they used a 24volt rectified, unsmoothed AC to power the tip to minimise the transient voltages you'd otherwise suffer by interrupting a smoothed 24v DC supply as per these KSGER stations. It seems they missed a trick in minimising electrolytic corrosion due to residual moisture issues by not rectifying the ac voltage - perhaps they do precondition their T12/T15 tips on a 12vac supply after all as part of the final bake out/quality control testing which the clone tip manufacturers don't bother with, preferring to leave all that to the hapless end user.

The moral of this being if you're going to use fake Hakko tips buy them in some quantity that equates to the cost of a single genuine Hakko tip, say a pack of ten or more. >:D That way, you won't be cursed by the "Sample of only one" effect in regard of the faulty/poor quality tip statistics figures. Well, it's a choice! ::)

JBG
 

Online KL27x

  • Super Contributor
  • ***
  • Posts: 3614
  • Country: us
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #82 on: September 25, 2019, 01:53:08 am »
Quote
I was thinking that this could probably be done with a modern cmos Z80 cpu, say one clocked at a modest 8MHz. 8 bit unsigned integer could cover the range 150 to 400 deg C where you'd assume an overflow (carry result) to simply indicate "Too bloody hot! Cut heater power, buddy!" and a zero would indicate too low a temperature to solder with.
Quote
You could equate each integer value (0 to 255) to 2 deg C (a one degree C accuracy is way more than we actually need - 5 deg C increments would be accurate enough for this task)
which is enough to go from 100 to 610 deg C.  :)

Yes, you can get 256 resolution from a byte, and that is plenty. AFTER you figure out the temperature of the tip. To do that within 2 degree accuracy will require WAY MORE resolution than 8 bits, because the data you are getting this from will have a huge temp swing from min to max, but the differences of heater/sensor temp (within a short enough period of time after turning off the heater to be practical) between say 300C while the iron is "floating" and 250C while floating will be very very small compared that very huge range you have to cover. That difference will not be 50 degrees at the sensor at Tmeasure. It will be a tiny fraction of 50C.* This is what you need the high resolution for. The temp of the heater/sensor rounds off to "really hot and barely relatable to tip temp" until you dissect it into very small bites (at least, best case is 25 smaller bites) to end up with actual 2C temperature resolution for the actual tip temperature.

So you need only a resolution of 25? No, you have to still cover the min max of the entire temp range, heater on or off. You have to cover that entire range with this high resolution net. But I said the sensor will be "hot" and differences are very small, so why the big range? Well, you also have to be able to measure temp when the heater has been off for awhile, say after reducing temp setting and it is falling. And there, the min just went low, real low.

I'm sure STM has 12bit ADC at least.

All of the stuff described from my "resolution post" would be done JUST to derive the actual tip temperature. That is all step 1. This is just to uncouple the sensor data from the heater.

*On second thought that might be a bad example. After reaching equilibrium that might be close to 50C. Here's a better example. The user is pressing the tip against the board, sinking out heat. This drops the temp of the base of the tip around the sensor. If you recorded at only 1 time after cutting off the heater, the reading will hit the set temp while the base is still below set temp. Because the differential between heater temp and tip temp will grow in that imperfect coupling between the two. So the actual tip temp will be lower than the station thinks it is. According to the single measurement, you're at the set temp; station will allow this sag. To correct for this problem is one of the reasons you need to take a second temp reading a discrete amount of time later. Here, the temperature of the sensor will be very slightly lower due to faster heat transfer vs the one floating at actual correct set temp. This is how you get rid of the sag, and why you need such a high resolution.
« Last Edit: September 25, 2019, 03:08:46 am by KL27x »
 

Online KL27x

  • Super Contributor
  • ***
  • Posts: 3614
  • Country: us
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #83 on: September 25, 2019, 06:14:07 pm »
This is why you need relatively high resolution.

This graph diplays temperature of the sensor over time, with T0 being when the heater is shut off.

In these two plots, one is under no thermal load. The other is while under a high thermal load.
In either case, the heater has been full duty cycle long enough to reach equilibrium. In either case, the temp of the tip is rising, but all the slack has been taken out of the thermal couplings by this time (and that's another thing the station/micro has to keep track of). It's just that at this particular time of sensor test, they happen to both have the same temp at the first reading near to T0. By resolving the tiny difference at Tmeasure, you can tell the true temp of the tip (or the future temp of the sensor).

The longer the heater is off, the closer it will get to the temp of the tip. But we don't want to wait that long.* If the temp is determined to still be under set temp, the heater is going back on at Tmeasure. The station won't get to see the rest of this graph. It makes the decision at Tmeasure.

This is why it needs relatively high resolution. The difference between tip temp of these two plots might be tens of degrees C (the two dotted horizontal lines are true tip temp). The difference could be, say, 50C under a very high load. But the difference in temp of the sensor at Tmeasure between the two could be a single degree. That single degree has to be resolved in (at least) 25 pieces to get 2C resolution of tip temperature. That's if those bites work out to near 2C all the way through that range, which it will probably not. That might result in 1C increments at one end of the range and 4C increments at the other. So you probably need at least double that resolution, say 0.02C increments of sensor temp to get 2C increments, and this would still include massive "pixelation." These are just made up numbers to illustrate the point, but you see the more resolution you have, the better your accuracy will be across the entire range. Pace ADS. 17 bit ADC. Starting to make sense, yet?

For the sake of simplicity, the true temp of the tip is considered static. Of course it will also start to drop at some point shortly after T0. But compared to the rate of change of sensor temp between T0 and Tmeasure, I think it's ok to ignore that across the time frame of interest. In this case the tip is the dog, and the heater/sensor is the tail.
 
*You want probably want multiple senses per second to be really responsive. Say you want a leisurely 10 reads per second. This makes each period 100ms. During the read, the heater is off. So if you wanted to be able to maintain at least 50% duty cycle (most T12 clones seem to reach higher), then the max your Tmeasure can be is 50ms. At 90% duty cycles, 10 times per second, the max your Tmeasure can be is 10ms. So we're at 10 samples per second, right now, and how much faster do you think you can get it? An analog iron with a proper sensor setup has an infinite sample rate. So much for the "faster response time" T12 fanboi argument. :) The higher the sample frequency you want to attain, the shorter Tmeasure. And the more ADC resolution you will need. Clock stability is also critical. This should benefit appreciably from an external crystal clock. If your clone runs on an internal ceramic resonator and uses a 12 bit ADC... meh. i don't think this is gonna be very accurate and responsive, if it attempts to correct for this error/sag, at all.


The top dotted line is the true temperature of the tip under no load (or load of ambient air, to be exact).
The bottom dotted line is the true temperature of same/similar tip under a heavy load.
The reason the sensor is at the same temp approaching T0 is because a differential between the temp of the heater/sensor and copper develops under load. The greater the heatsink/load on the tip, the greater this differential will get. This is the source of the error, resulting in sag/droop, if you do not correct it. This represents, say, the maximum amount of error you might need to correct. The "worst case" scenario. Beyond that amount, the heater wouldn't be able to keep up, anyway, and will be pegged on full in either case, with or without correction.
« Last Edit: September 26, 2019, 08:13:59 pm by KL27x »
 

Offline Johnny B Good

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: gb
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #84 on: September 26, 2019, 11:48:26 pm »
 Apologies for the delay in replying. Quite apart from the need to chew over what you were saying, "Real Life" concerns had also intruded into my thoughts. I'm not talking about the more mundane RL concerns of the great unwashed you understand, more those of the, nevertheless still mundane, concerns of dealing with the results of shopping (on line - is there any other form of shopping for those of us now stuck between a rock and a hard place when it comes to buying electronic components and such like items?). I won't go into details here but, if you're interested, I've added them as a footnote[1] for you to peruse at your leisure. I've little doubt that I'm merely adding to the shared experience of the collective so make of it what you will. ::)

 My thoughts about using a Z80 cpu and 8 bit ADC data were more a vague 'thought experiment" to give the use of the much better endowed and orders of magnitude faster STM32 micro controller actually used in these KSGER soldering stations some context.

 I do see the point you're making about using higher resolution data from the tip sensor ADC in being able to more swiftly detect changes around the inflection points of the temperature curves in order to more precisely spot the trends and initiate the correct response in a more timely fashion to minimise the effect of lag.

 I suspect being able to differentiate between temperatures separated by only 0.02 deg C is likely to be thwarted by noise in the signal which will need to be filtered out (in the digital domain rather than at the analogue input stage).

 Luckily, most of this noise will be relatively high frequency and a low bandwidth of, say, 500Hz or less would likely serve our needs quite nicely in this application. Plenty of time for an STM32 to do the necessary DSP sufficient to perhaps resolve to within half a degree's worth of accuracy which I feel may well be 'overkill' even in this case.

 However, like you, I don't have access to any hard data to back up such an opinion. You could well be right in assuming Pace have a valid justification for the use of a 17 bit ADC in their very expensive soldering station kit.

 Inspired by that other EEVBlog thread (the OP actually), I soldered an earth strap between the TPHs used by the encoder's mounting tabs to an actual ground connection nearby to address possible ESD issues and added a BY198 diode (in place of the RS1M diode suggested by Floobydust) across the heater connections to ground clamp the flyback pulse produced when the power is cut off to the 4μH's worth of inductance in the 8 ohm heating element and cabling in order to protect both the switching MOSFET and thermocouple amplifier.

 Obviously, I was anxious to test out these modifications so plugged it into a repurposed block filter plug adapter which allows connection of an analogue wattmeter so I could immediately see any possible power up overloading effect I may have caused as a result of these modifications.

 The point of mentioning all of this is that I could see that the soldering station was applying bursts of power after drawing a steady 80 watts off the mains supply during the initial 4 or 5 seconds heating time to 300 deg phase at the rate of about five to six times per second.

 At this 300 deg setting, the critically damped Metrawatt meter movement integrates this pulsating power draw to a reading of 10 watts with a small +/- 2W wobble imposed by the pulsations. Meanwhile, the KSGER indicates a percentage of applied power figure of 7 or 8 percent which seems reasonable enough considering the different points of measurement involved.

 Thankfully, I had managed to avoid doing more harm than good (hopefully nothing but good) in making these modifications. It would seem that your 'guesstimate' of a "leisurely 10 reads per second" is pretty well right in the ballpark (if a little less "leisurely" than the update rate seemingly used by the KSGER STM32 control algorithm).

 In view of the 0.5μs time constant (worst case whilst the extra clamp diode I've added to complement the body diode in the switching mosfet is handling the flyback pulse current), it strikes me that a 10μs delay from heater current switch off should leave the subsequent sampling of the TC voltage completely unmolested.

 A further 31 samples at a modest 50KS/s could be collected over a total off time of just 0.64ms to feed the DSP noise filter algorithm to refine our TC voltage signal so it looks like we needn't have to shut the power off for longer than 1ms in every 100 (or 200 in the case of the KSGER) when ramping the temperature up or maintaining tip temperature with the heaviest duty tip against a heavy duty heatsink load.

 In practice, it would seem that even when soldering a 'difficult joint' with the BC3 tip (the most substantial of my bargain pack of ten clone tips) at a 350 deg set temperature, it seems unlikely that the average power percentage will need to go above the 40% mark.

 JOOI, I've actually tested that just now with a British two pence coin which I'd previously 'tinned' a few weeks back. It took several seconds to establish a good thermal contact with some fresh multicore solder before the percentage maxed out around the 40% mark.

 By blowing hard onto the penny and solder tip, I did manage to raise that to 48% briefly, noticing that the reading on my wattmeter had reached the thirty watt mark as I raised my head to look at the meter. Also noted was the modest 3 or 4 degree sag on the station's indicated tip temperature during all of my 'huffing and puffing' and the more interesting rebound (overshoot if you prefer) to 360 deg before dropping back to the set temperature upon removing the tip from contact with my "difficult test joint".

 All in all, I thought this was an excellent result despite not complicating the test with actual tip temperature readings using my FG-100 tip thermometer. However, though such testing might leave me a little less impressed, the best I could do would be to read the unloaded tip temperature. Ideally, I'd need a remote thermocouple that could be applied to the tip and my 'test joint' to get a more accurate assessment of the actual temperatures being achieved in such use.

 Since none of my existing collection of DMMs sport such a K type TC probe port, I'm considering buying a K type TC probe which I can connect to the FG-100's TC terminals to read the TC voltage and hence give me a temperature readout. As far as I'm aware, the clip in TCs used by these meters are just a K type TC in a compact form so I should be able to save the expense of buying yet another DMM just for the sake of a K type TC port. If the readings from a remote K type TC agree closely enough, it might be worth modifying the FG-100 to allow a standard K type TC plug ended temperature sensor to be be plugged into it (after removing the clip in TC, of course) for any such future tests.

 The other observation I've taken note of since doing the clamp diode modification to the heater circuit (and possibly even before) is the elimination of the "New Tip Instability" issue. It seems my sporadic use of the BC3 tip over the past week or so has successfully completed the "Burn In" process, rendering it entirely stable even right up to the unboosted 450 deg limit (in theory, this could be boosted another 100 deg but I have no desire to subject my BC3 tip to any such further abuse).

 Assuming that all such clone tips will settle down to a more civilised behaviour with this KSGER station after a few heating and cooling cycles of use, I'm tempted to put them three at a time in series with a 36v 400VA transformer to dry them out "en masse" without the possible detriment of using DC bias which could activate a (seemingly) temporary electrochemical reaction during the initial drying out phase. If my hypothesis about residual damp and electrochemical reactions is correct, I'd like to minimise any such electrolysis effects even if they do appear to be only of a transient nature.

 A basic 12v 24VA transformer based "Tip Conditioner" could become a 'must have' KSGER soldering station accessory for the more prolific user of clone T12 tips. >:D Whether such an accessory would be of any use to someone who sticks with the more pricey Hakko 'originals' remains an open question. However, in view of Hakko's use of rectified (unsmoothed) AC to power the tip heaters in the FX-951 soldering stations and the absence of any mention of a need with brand new Hakko tips to do a similar burning in exercise, I suspect not.

 In any case, the benefit of a 12v transformer based new clone T12 tip conditioning 'gadget' is an, as yet, untested idea of mine. Even if it does solve the "New Tip Syndrome", it's hardly a marketable good since anyone in the position of using a KSGER or copy T12 soldering station with clone T12 tips will be looking to minimising their spending and, worst still from a marketeer's point of view, more than capable of repurposing a suitable mains transformer from their, no doubt burgeoning pile, of "salvage' to such a basic function (assuming that such a gadget is even desired in the first place).

 If this scheme to condition new clone tips with a 12v transformer proves successful, I may well decide to "sell my soul" to Youtube to sign up as a contributor to publicise my "Wizdumb" to the greater world (small as this section of it is) at large. :)

[Notes]

[1] I've already mentioned that broken Mustool G600 LCD microscope I'd received from Banggood a couple of days ago. Well, the latest upshot to that is that it looks like they want to offer me a refund of its cost (but not the 66 pence shipping insurance) with no mention of what I'm supposed to actually do with the broken unit.

 It looks like they don't want me to return it which I suspect is just the economics of the situation (I had suggested quite strongly in my first and previous reply that I didn't want to accept their offer of a replacement if they were expecting me to pay the cost of the return shipping after advising them that a local repair would not be economically viable - their other suggestion).

 This afternoon's message from them posed the offer of the refund as a question sans question mark, so I asked for clarification about whether they wanted me to keep hold of the original for their courier to collect or whether they just wished me to dispose of it as I saw fit, also pointing out the fact that they'd left the shipping insurance cost out of their refund calculation. Just for good measure, I asked them to unambiguously state their position with regard to the return, or not, of the broken microscope and my acceptance of their refund offer.

 I'm not too bothered about the 66p shipping insurance charge since, following an Ebay breadcrumb trail search, I managed not only to track down a cheaper version with the grotty plastic suction windscreen stand from a seller shipping from within the UK, I managed to find another who was offering the later version with the sturdy aluminium adjustable base for less than a quid more, making the whole deal (the net refund plus cost of the uk sourced replacement) some 47 pence cheaper than the original Banggood price plus shipping insurance charge.

 Now, all that remains is for Banggood to confirm the exact details of their refund offer - if they want the damn thing back at their cost, they can have it back - I don't need it any more for its base as an upgrade to that first cheap Ebay find. Coincidentally, I've just this moment (22:23 BST) received a second email from the ebay seller from whom I'd purchased the best of those two Mustool G600 microscope kits earlier this afternoon to inform me that it's now being posted which bodes well for an earlier delivery than the promised Thursday the 3rd of October date.

 Past experience with ebay sellers shipping out of UK warehouses has been quite good in regard of timely deliveries. It's been rare indeed for a delivery to over run its promised date of delivery (typically a week after placement of the order) by more than a day or two with most arriving within just two to four days. This is a far cry from the Banggood experience of late (only my very first Banggood order (for that KSGER and some useful accessories btw) managed to arrive within just 8 days of being ordered).

 You might think I'd have had enough by now what with a large order that I'd placed on 11th Sep still stuck at the processing stage with no indication as to why it is now four days past its promised shipping date other than to a query I'd slipped into my first reply to their initial response to my reporting the broken microscope kit.

 It appears that the problem is "out of stock" ("What! All eight items? Are you serious?" being effectively my response to that information, along with a request for more details so I can decide how best to deal with it). TBH, I'm surprised they hadn't done what they'd done with another large order where they'd split it up, presumably to get what items were in stock shipped out to prevent the out of stock items needlessly holding up the whole order - a point I'd raised in my response to this news.

 It will be interesting to say the least as to how they'll respond to my utter disappointment in their handling of this "out of stock" issue. TBH, I can afford to wait a little longer on this order since nothing is urgently needed, not even the 30V10A 4 digit LED display bench supply headline item that I managed to snag for just £37.82 At that price, some ten quid cheaper than even the cheapest ebay/amazon seller alternatives, I can well afford to bide my time.

 As I was saying, you might think I'd have had enough of my dealings with Banggood but as it happens, I only went and placed an order for 3 metres of flexible 5 core silicone cable and a DIY kit KSGER FX9501 T12 Soldering Handle for just £6.71.

 Again this was an item I had been searching for during the past week or so on Amazon and Ebay only to discover silly prices for the ones with a 4 core cable and the stupid 8 pin din plug intended for the Hakko soldering stations, typically priced at 11 quid and up for something I'd have to spend yet more money on. Searching directly on Banggood's web site had failed to unearth where this little gem had been hidden (it even includes the GX12-5 connectors (both the panel and the handle connectors - Yay!) and it was not until I tried a duckduckgo search with the inclusion of they key word "DIY" that I finally discovered where BG had hidden it.

 No cable supplied though, hence the purchase of 3 metres for just £4.14 for me to split into 1.5 metre lengths for the spare and to upgrade the 1 metre cable currently attached to my existing iron handle. I much prefer a DIY assemble it yourself kit over the ready assembled units simply because it saves the time you'd spend taking such ready made handles apart in order to fettle the typically piss poor quality of assembly seen in these items of 'best Chinese manufacture'. This way, you'll know exactly how good the quality of assembly will be before using it in anger. :)

 In spite of several reviewers' prejudices against its cheap "plasticy" construction, I was surprised at how difficult it had proved to be to locate one at the hoped for 'cheap' price point. I had deliberately chosen this particular KSGER T12 soldering station package for this soldering handle, seeing through the reviewers' prejudices to recognise the charms of its quick tip change, light weight and short tip to grip distance with the latter being the overriding charm of its construction in view of its intended purpose of soldering smd components onto cramped circuit boards where the last thing you need is a hulking great tip retention collect to get in the way of a long distance tip. Quite frankly, I'm rather relieved that I was finally able to track down and order another one so cheaply as a spare. :)

JBG
 

Online KL27x

  • Super Contributor
  • ***
  • Posts: 3614
  • Country: us
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #85 on: September 30, 2019, 10:13:29 pm »
Quote
In practice, it would seem that even when soldering a 'difficult joint' with the BC3 tip (the most substantial of my bargain pack of ten clone tips) at a 350 deg set temperature, it seems unlikely that the average power percentage will need to go above the 40% mark.

 JOOI, I've actually tested that just now with a British two pence coin which I'd previously 'tinned' a few weeks back. It took several seconds to establish a good thermal contact with some fresh multicore solder before the percentage maxed out around the 40% mark.

 By blowing hard onto the penny and solder tip, I did manage to raise that to 48% briefly, noticing that the reading on my wattmeter had reached the thirty watt mark as I raised my head to look at the meter. Also noted was the modest 3 or 4 degree sag on the station's indicated tip temperature during all of my 'huffing and puffing' and the more interesting rebound (overshoot if you prefer) to 360 deg before dropping back to the set temperature upon removing the tip from contact with my "difficult test joint".

All this stuff is fine and all, but you can't rely on the temperature readout of the station. W/e you measured is meaningless, because we already know 350C >> 200C, and 200C is enough to melt solder. The point of everything I mentioned about sensor/heater coupling means that the station does not necessarily know the correct temp of the tip. You have to measure the temp of the joint while you're doing it, using an accurate temp probe. By putting a temp testing thermocouple, for instance, between the tip and the 2 pence coin.

When the station display reads 3-4 degrees low, the joint might really be 40-50C low (Maybe 25C lower than a "good iron" under this same test, because both will sag between base of tip and the point of the tip that 25C. But the T12 clone WILL have additional sag from the sensor coupling, unless it is able to correct for this). These numbers are just thrown out and will vary by size of heatsink. Greater the heatsink, greater the sag. When it "overshoots up to 360" it is likely still lower than 350, and it might still be 30-40C low. This is my experience/finding with cheap T12 clones, and I suspect there are more than just the two varieties of clones I have experience with which don't bother to try to correct for this sag/error.

So just because the iron doesn't need to go above 40% duty cycle to melt solder on to the coin doesn't mean the temp isn't sagging. (If you use a firestick that is controlled by a dimmer, it can also melt solder onto stuff at a reduced duty cycle that hits 350C at no load). IOW, just cuz it's only at 40% doesn't mean that you can solder to a larger heatsink without bumping up the set temp. It's not like the iron will necessarily bump up duty cycle to 80% in order to maintain the same temperature at the base of the tip on this larger coin. It could bump up duty cycle to 70% and allow the base of the tip to sag even lower from the set point under no load. If the error is not corrected, the temp will sag in relation to size of the heatsink; that's how the coupling error is expected to behave. The display will show artificially high numbers and obviously cut power because it assumes it is at set point. And the iron that corrects for this sag would go to a higher duty cycle and keep the joint closer to the no-load set temp. IOW, if you turn down the set temp on both irons (and actually measure set temp not trust the display, of course), the one with worse or nonexistent temperature correction is going to stop being able to solder to that 2 pence coin, first, despite set temp (under no load) still being significantly higher than 200C and identical to the better iron which is still able to solder to that coin.

Because all irons sag between the base of the tip (where the sensor is) and the point of the tip under load, this can't easily be removed from the reading. You have to compare two irons and they both have this sag. This is why you must use similar size and shape tip, to that effect, so that the temp sag between base of the tip and the point is consistent. And of course you have to choose test joint/sink which is not so big that the iron cannot reach its (good or flawed) set point under even max duty cycle. (Fact that the iron does not go above 40% duty cycle is the proof that this sag is due to control circuit not maxed out on power; the station COULD do better if it had better sensor data/handling). Just purely a guess, cuz I don't know what a 2 pence coin looks like, but I would guess with the BC3 tip, my own clones might start to fail at around a true set temp of 300-320ish whereas the 888 would keep trucking no problem.

The T12 BC3 tip has every advantage over the T18 C3 tip regarding sag between the base of the tip and the point. It is shorter and the tip tapers down to 3mm bevel. The T18 is about twice as long and is closer to 3mm cylinder the entire length between base and tip; it doesn't have the fat taper to the thicker base. (This is what the "B" in "BC" means; it means that it's a hybrid bevel/conical... a conical base that ends in a bevel at the tip.) And Hakko 888 C3 tip significantly beats clones using BC3 in this test. So it is easy to see that there is additional sag due to heater/sensor coupling in some T12 clones.

Ideally, the test joint will be purely a heatsink and it should not require necessarily a heavy thermal mass/inertia of its own. A blank piece of copper clad is more ideal than a 2 pence coin. To get a meaningful result, you should wait, in either case, until the heatsink has attained a stable temperature and no longer changes. The sheet of copper clad will be a bit more stable. The pence coin will be more sensitive to ambient air and convention currents and w/e heatsink you're setting it on and pressure between the two.

I suspect many clones fail to correct for the coupling error. But because you describe behavior that hints at only a super slow 5 reads per second, I am hoping that this KSGER version actually attempts to do this. I might have to buy one to test just for the hell of it.
« Last Edit: October 01, 2019, 12:23:14 am by KL27x »
 

Online KL27x

  • Super Contributor
  • ***
  • Posts: 3614
  • Country: us
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #86 on: October 01, 2019, 08:41:54 pm »
You mentioned earlier about moving the tip of the thermocouple heater/sensor away from the rest of the main coil, in order to reduce this coupling effect. I wonder what happens when you put an intermediary conductor between the two halves of the thermocouple. This should stop the effect, maybe?

So you have your heater coil, and the two metals don't connect, directly. There is an intermediary material that electrically connects them. Then there's just two tiny gauge wires of the thermistor metals that connect to the heater wires, directly, bypassing the intermediary conductor. The ends of these two tiny sensor wires are directly connected and placed a tad away from the heater proper, perhaps a bit closer to the tip.

I still think error correction will improve even this setup, though. So I doubt it would be worth the cost IF error correction can be done well.

Hum, naw, that doesn't work. The intermediary conductor will short the voltage produced at the tip.
« Last Edit: October 01, 2019, 08:43:45 pm by KL27x »
 

Offline Johnny B Good

  • Frequent Contributor
  • **
  • Posts: 256
  • Country: gb
Re: Hakko still the best option for a good quality hobbyist soldering station?
« Reply #87 on: October 03, 2019, 03:08:28 am »
 Just keep in mind that the thermocouple effect arises out of the temperature gradient between the hot and the cold junctions (Seebeck effect). No voltage is actually produced at the junctions themselves.

 I must admit that until I read that wikipedia article on thermocouples, I had assumed the voltage was produced at the junctions. It turns out that you can even connect the thermocouple alloy wires via a third metallic alloy, say a crimp, and not have it effect the voltage generated by the Seebeck effect provided there is no temperature difference between the thermal wire junction connections to the bridging crimp.

 Ignoring the ageing effect for the moment, it doesn't matter that one or both thermocouple wires run at an elevated temperature along their length compared to the hot end junction. The voltage produced is purely defined by the difference in temperature between the hot (measuring) and the cold (reference) junctions.

 Unfortunately, most base metal alloys used for the cheaper, less exotic TCs suffer an ageing effect which typically results in a reduction of sensitivity and is aggravated by running the wires through an elevated temperature region such as the heater portion of such a combined heater and TC direct drive cartridge. Hakko have obviously chosen the most economically priced alloys used to achieve virtually no such ageing effect, accepting the halving of sensitivity as the price for such stability.

 One might assume that the clone tips use the same combination of alloys to maintain compatibility in regard of the TC sensitivity but there may be other cheaper alternatives that could match the ~20 microvolts per K sensitivity of the Hakko tips so only time would tell whether they've used the same alloys and hence have the same agelessness of the Hakko originals.

 If it turns out that Hakko had chosen the cheapest possible solution, this is unlikely to be a factor in the quality of the clone/fake tips (but that still leaves many more ways for these cheap clones to fall far short of the quality of an original Hakko tip).

 Regardless of the exact mechanism involved in the NTS effect with these clone tips (galvanic or possibly a pre-compensation effect for a thermocouple ageing process), there still remains the option to use Hakko original tips making these cheap STM32 based soldering stations worth taking a punt on for hobbyist level use if you're prepared to remedy their blatant safety issues and the poor quality of assembly of the supplied soldering handles.

JBG
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf