Unable to calibrate VPP

[AJROMR]

I have built v3.1a pcb using SMT components, a good MT3608 module but I am still having issues with afterburner calibration.

In order to get detailed info I have made the following modifications to the sketch files:

---

afterburner.ino (Line# 2703):
if (offset >=0 && offset <= 9) {

aftb_vpp.h (Line# 41)
#define VPP_VERBOSE 1

---

I have tested it with two different Arduino UNO R3 (chinese clones) and as I was not able to find the issue I have built the following circuit in a breadboard but the results were very similar.

There is only one change, in the v3.1a pcb I used a MCP4131-103 digital pot, in this breadboard version I used a DIP-8 MCP4151-103 digital pot.

v3.1a pcb (SMT) calibration output:

---

./afterburner s

Turn the Pot on the MT3608 module to set the VPP to 16.5V (+/- 0.05V)

varSetVppIndex 64
varSetVppIndex 112
varSetVppIndex 124
varSetVppIndex 126
varSetVppIndex 128
16.52
16.52
16.52
16.52
VPP OK
varSetVppIndex 0

---

./afterburner b

Using cal offset: 25
VPP voltages are scanned - this might take a while...

VPP calib. offset: 25
varSetVppIndex 1
varSetVppIndex 1

1. varSetVppIndex 2
2. *Index for VPP 900 is 1
   vppIndex 1
   varSetVppIndex 1
3. varSetVppIndex 2
4. *Index for VPP 950 is 1
   vppIndex 2
   varSetVppIndex 1
5. varSetVppIndex 2
6. *Index for VPP 1000 is 1
   vppIndex 3
   varSetVppIndex 1
7. varSetVppIndex 2
8. *Index for VPP 1050 is 1
   vppIndex 4
   varSetVppIndex 1
9. varSetVppIndex 2
10. *Index for VPP 1100 is 1
    vppIndex 5
    varSetVppIndex 1
11. varSetVppIndex 2
12. *Index for VPP 1150 is 1
    vppIndex 6
    varSetVppIndex 1
13. varSetVppIndex 2
14. *Index for VPP 1200 is 1
    vppIndex 7
    varSetVppIndex 1
15. varSetVppIndex 2
16. *Index for VPP 1250 is 1
    vppIndex 8
    varSetVppIndex 1
17. varSetVppIndex 2
18. *Index for VPP 1300 is 1
    vppIndex 9
    varSetVppIndex 1
19. varSetVppIndex 2
20. varSetVppIndex 3
21. varSetVppIndex 4
22. varSetVppIndex 5
23. *Index for VPP 1350 is 4
    vppIndex 10
    varSetVppIndex 4
24. varSetVppIndex 5
25. *Index for VPP 1400 is 4
    vppIndex 11
    varSetVppIndex 4
26. varSetVppIndex 5

---

./afterburner m

VPP calib. offset: 25
VPP: 4.2 - 5.0V : varSetVppIndex 0
11.25
VPP: 9.0V : varSetVpp 0:1
varSetVppIndex 1
12.95
VPP: 12.0V : varSetVpp 6:1
varSetVppIndex 1
12.94
VPP: 14.0V : varSetVpp 10:4
varSetVppIndex 0
varSetVppIndex 4
12.97
VPP: 16.0V : varSetVpp 14:121
varSetVppIndex 58
varSetVppIndex 82
varSetVppIndex 94
varSetVppIndex 100
varSetVppIndex 103
varSetVppIndex 104
varSetVppIndex 105
varSetVppIndex 106
varSetVppIndex 107
varSetVppIndex 108
varSetVppIndex 109
varSetVppIndex 110
varSetVppIndex 111
varSetVppIndex 112
varSetVppIndex 113
varSetVppIndex 114
varSetVppIndex 115
varSetVppIndex 116
varSetVppIndex 117
varSetVppIndex 118
varSetVppIndex 119
varSetVppIndex 120
varSetVppIndex 121
15.98

---

Breadboard calibration output:

---

./afterburner s

Turn the Pot on the MT3608 module to set the VPP to 16.5V (+/- 0.05V)

varSetVppIndex 64
varSetVppIndex 112
varSetVppIndex 124
varSetVppIndex 126
varSetVppIndex 128
16.48
16.50
16.52
16.49
VPP OK
varSetVppIndex 0

---

./afterburner b

Using cal offset: 25
VPP voltages are scanned - this might take a while...

VPP calib. offset: 25
varSetVppIndex 1
varSetVppIndex 1

1. varSetVppIndex 2
2. varSetVppIndex 3
3. *Index for VPP 900 is 2
   vppIndex 1
   varSetVppIndex 2
4. varSetVppIndex 3
5. *Index for VPP 950 is 2
   vppIndex 2
   varSetVppIndex 2
6. varSetVppIndex 3
7. *Index for VPP 1000 is 2
   vppIndex 3
   varSetVppIndex 2
8. varSetVppIndex 3
9. *Index for VPP 1050 is 2
   vppIndex 4
   varSetVppIndex 2
10. varSetVppIndex 3
11. *Index for VPP 1100 is 2
    vppIndex 5
    varSetVppIndex 2
12. varSetVppIndex 3
13. *Index for VPP 1150 is 2
    vppIndex 6
    varSetVppIndex 2
14. varSetVppIndex 3
15. varSetVppIndex 4
16. *Index for VPP 1200 is 3
    vppIndex 7
    varSetVppIndex 3
17. varSetVppIndex 4
18. varSetVppIndex 5
19. *Index for VPP 1250 is 4
    vppIndex 8
    varSetVppIndex 4
20. varSetVppIndex 5
21. *Index for VPP 1300 is 4
    vppIndex 9
    varSetVppIndex 4
22. varSetVppIndex 5
23. *Index for VPP 1350 is 4
    vppIndex 10
    varSetVppIndex 4
24. varSetVppIndex 5
25. varSetVppIndex 6
26. *Index for VPP 1400 is 5
    vppIndex 11

---

./afterburner m

VPP calib. offset: 25
VPP: 4.2 - 5.0V : varSetVppIndex 0
11.39
VPP: 9.0V : varSetVpp 0:2
varSetVppIndex 2
13.04
VPP: 12.0V : varSetVpp 6:3
varSetVppIndex 3
13.04
VPP: 14.0V : varSetVpp 10:5
varSetVppIndex 0
varSetVppIndex 5
13.06
VPP: 16.0V : varSetVpp 14:10
varSetVppIndex 3
varSetVppIndex 10
13.11

---

On the other hand,

1. README.md: There are several mentions to R8 but I think you want to mean R9.
2. bom.txt: "(R1 - 5x)" but in the pcb there are only 4 footprints for R1.

Thanks in advance

[ole00]

Hi,

sorry to read about the calibration issue.
few observations:

1. the MT3608 does look OK to me - like the type that should work fine
2. Arduino UNO Chinese clones - they usually work OK, I find them slightly better that genuine Arduino UNO
3. R8 /R9 mismatch and BOM - yes, on recent PCB it is R9 POT. I'll fix the description, thanks.
4. MCP4131 and MCP4151 is autodetected in Arduino software, both should work OK.
5. when using './afterburner s' you did not mention whether you measured the output of the MT3608 module by a multimeter. It is recommended, to verify the reading from Arduino matches real value.
6. when using './afterburner b' I can see the issue : your listing prints '*Index for VPP 900 is 1'. The value must be between 15 and 35. Value 1 is definitely a wrong result. That either means the R9 pot is set to an incorrect extreme, or the Arduino ADC reading VPP_SENSE signal reads a wrong value. Double check that the R5 (100k) and R6 (20k) on the voltage divider are the correct values on your board. Also, double check that R2 is 3k3 on your board.

So, ensure the multimeter reading (point 5) matches the print out from Arduino. Then double check the resistors (point 6)
and if it is OK then rotate R9 pot to extremes and repeat './afterburner s' and './afterburner b'. If you rotate Pot R9
you must re-do the './fterburner s' and make sure the voltage output is 16.5V (also on multimeter). Only then you do './afterburner s'. Check the listing '*Index for VPP 900 is [X]'in both cases (R9 extremes). The [X] should really be different for both extremes.
Let me know how it goes.

[ole00]

Just for a reference, here is my printout when running ./afterburner b
ER: cal offset failed
VPP voltages are scanned - this might take a while...

VPP calib. offset: 15
*Index for VPP 900 is 15
*Index for VPP 950 is 34
*Index for VPP 1000 is 49
*Index for VPP 1050 is 61
*Index for VPP 1100 is 72
*Index for VPP 1150 is 81
*Index for VPP 1200 is 88
*Index for VPP 1250 is 95
*Index for VPP 1300 is 101
*Index for VPP 1350 is 106
*Index for VPP 1400 is 111
*Index for VPP 1450 is 115
*Index for VPP 1500 is 118
*Index for VPP 1550 is 122
*Index for VPP 1600 is 125
*Index for VPP 1650 is 128
Calibration OK

[ole00]

I've printed the calibration when Pot R9 is in the extremes. It looks quite different from what you sent earlier:

Non optimal calibration: R9 rotated fully counter-clockwise:
VPP calib. offset: 15
*Index for VPP 900 is 1
*Index for VPP 950 is 1
*Index for VPP 1000 is 17
*Index for VPP 1050 is 37
*Index for VPP 1100 is 52
*Index for VPP 1150 is 65
*Index for VPP 1200 is 76
*Index for VPP 1250 is 86
*Index for VPP 1300 is 94
*Index for VPP 1350 is 101
*Index for VPP 1400 is 107
*Index for VPP 1450 is 112
*Index for VPP 1500 is 117
*Index for VPP 1550 is 121
*Index for VPP 1600 is 125
*Index for VPP 1650 is 128
Calibration OK

Non optimal calibration: R9 rotated fully clockwise:
VPP voltages are scanned - this might take a while...

VPP calib. offset: 15
*Index for VPP 900 is 73
*Index for VPP 950 is 80
*Index for VPP 1000 is 87
*Index for VPP 1050 is 92
*Index for VPP 1100 is 97
*Index for VPP 1150 is 102
*Index for VPP 1200 is 106
*Index for VPP 1250 is 109
*Index for VPP 1300 is 112
*Index for VPP 1350 is 115
*Index for VPP 1400 is 118
*Index for VPP 1450 is 120
*Index for VPP 1500 is 122
*Index for VPP 1550 is 124
*Index for VPP 1600 is 126
*Index for VPP 1650 is 128
Calibration OK

[ole00]

And perhaps a little explanation of what the Index [x] means. The Digi POT has 128 positions (or taps) and each position provides a different resistance. Each POT position (or resistance) gives a different output voltage on the MT3608 module. When running the 'b' command, the Arduino sets the POT to every position and records which position corresponds to each output voltage (VPP, which is sensed by Arduino's ADC) that is used to program the GAL chips. For example, to get VPP voltage to 16.5V the Pot must be set to position 128 (Index for VPP 1650 is 128). Of course, to make it working correctly the Arduino's ADC must read the VPP voltage right - that's why the multimeter verification is required.

[AJROMR]

Thank you very much for your reply.

1. the MT3608 does look OK to me - like the type that should work fine

I have 3 or 4 different types, all of them should be working according to your pics.

4. MCP4131 and MCP4151 is autodetected in Arduino software, both should work OK.

I used MCP4131 in SMT v3.1a board version and MCP4151 in the breadboard version (VPP circuit) but I am getting same results with both boards.

5. when using './afterburner s' you did not mention whether you measured the output of the MT3608 module by a multimeter. It is recommended, to verify the reading from Arduino matches real value.

Sorry, './afterburner s' measures "16.50" but the multimeter ~16.16VDC.

6. when using './afterburner b' I can see the issue : your listing prints '*Index for VPP 900 is 1'. The value must be between 15 and 35. Value 1 is definitely a wrong result.

I know but I am just getting 1.

That either means the R9 pot is set to an incorrect extreme

R9 is set exactly to 1kΩ (between pins 1 & 2).

Double check that the R5 (100k) and R6 (20k) on the voltage divider are the correct values on your board.

I have also measured/checked both resistor values using the TH drills.

(sorry about the flux, I have checked/reworked all the solderings several times)

Also, double check that R2 is 3k3 on your board.

The ohmeter measures 3301Ω between R2 pin (S) and R9 pin 2.

---

./afterburner s
Turn the Pot on the MT3608 module to set the VPP to 16.5V (+/- 0.05V)

16.50
16.50
16.51
16.50
VPP OK

---

./afterburner b
ER: cal offset failed
VPP voltages are scanned - this might take a while...

VPP calib. offset: 25
*Index for VPP 900 is 1
*Index for VPP 950 is 1
*Index for VPP 1000 is 1
*Index for VPP 1050 is 1
*Index for VPP 1100 is 1
*Index for VPP 1150 is 1
*Index for VPP 1200 is 1
*Index for VPP 1250 is 1
*Index for VPP 1300 is 10
*Index for VPP 1350 is 50
*Index for VPP 1400 is 75
*Index for VPP 1450 is 92
*Index for VPP 1500 is 105
*Index for VPP 1550 is 115
*Index for VPP 1600 is 123
*Index for VPP 1650 is 128
Calibration OK

---

I guess there is something wrong but I cannot figure what is happening. I'd appreciate any help. Thanks in advance.

[ole00]

Hi,

it looks like the issue is mt3608 module. I thought this type of the module is the older (good/working) type, but it looks like it is not. Just yesterday, another user @hubertushirsch reported the issue (#31 , bottom of the thread) with that type of the module. Apology for that, that was my mistake.

I think it might be possible to fix the mt3608 module without desoldering the potentiometer. I think a replacement of the 2k2 resistor (in between the pot and the capacitor) with a higher value should do the trick (perhaps a value like 20k or 30k), but I'd have to investigate.
BTW. I like the way you connected the module by using the headers.

[ole00]

I did a few tests. The new modules can be used be used without desoldering the pot by replacing the 2k2 resistor by 12k or 15k (possibly 18k) resistor. It is a simple mod: heat the soldering iron to 350 deg C, put a medium size solder blob on the tip of the soldering iron, touch the resistor (in the middle) for a second (it heats up and the solder pads get melted), then slide it off from the board by applying a slight pressure on the resistor. Soldering the new resistor is not hard, I tried a through hole resistor (when testing 20k resistor - did not give me a good result), 10k SMT resistor (wrong results) and then 12k SMT resistor (0603 package) which works OK (not ideal, but should do). I did not test 15K or 18k replacement resistors (I don't have them available now), but I'd reckon these should be fine.

The 12k resistor gives the following calibration values (with the new type of the mt3608 module):
*Index for VPP 900 is 11
*Index for VPP 950 is 32
*Index for VPP 1000 is 47
*Index for VPP 1050 is 60
*Index for VPP 1100 is 71
*Index for VPP 1150 is 80
*Index for VPP 1200 is 88
*Index for VPP 1250 is 95
*Index for VPP 1300 is 101
*Index for VPP 1350 is 106
*Index for VPP 1400 is 111
*Index for VPP 1450 is 115
*Index for VPP 1500 is 119
*Index for VPP 1550 is 122
*Index for VPP 1600 is 125
*Index for VPP 1650 is 128

The 12k resistor is the absolute minimum for Afterburner, when I tried 10k the 4.2 - 5V VPP gave me 5.66V which is out of spec. Do not use 10k resistor replacement. The 20k replacement should work, but the calibration values are less than ideal. I'd reckon the best option is to use 15k resistor replacement.
Hope it helps.

Edit: just to be clear which resistor needs to be replaced , here is the snapshot:

[AJROMR]

it looks like the issue is mt3608 module. I thought this type of the module is the older (good/working) type, but it looks like it is not. Just yesterday, another user @hubertushirsch reported the issue (#31 , bottom of the thread) with that type of the module. Apology for that, that was my mistake.

No problem at all. During the last 3 years I have bought several dozen of them and have not been able to find one that works properly with this pcb.

BTW. I like the way you connected the module by using the headers.

Thank you. This makes it easier to test one MT3608 module after another. It would be better if the board had an extra FB pad just below the MT3608's FB pin.

I did a few tests. The new modules can be used be used without desoldering the pot by replacing the 2k2 resistor by 12k or 15k (possibly 18k) resistor.

That did the trick!

---

./afterburner s
Turn the Pot on the MT3608 module to set the VPP to 16.5V (+/- 0.05V)

16.50
16.50
16.50
16.51
VPP OK

---

./afterburner b
ER: cal offset failed
VPP voltages are scanned - this might take a while...

VPP calib. offset: 25
*Index for VPP 900 is 25
*Index for VPP 950 is 38
*Index for VPP 1000 is 50
*Index for VPP 1050 is 59
*Index for VPP 1100 is 67
*Index for VPP 1150 is 74
*Index for VPP 1200 is 81
*Index for VPP 1250 is 86
*Index for VPP 1300 is 91
*Index for VPP 1350 is 95
*Index for VPP 1400 is 99
*Index for VPP 1450 is 103
*Index for VPP 1500 is 106
*Index for VPP 1550 is 109
*Index for VPP 1600 is 112
*Index for VPP 1650 is 114
Calibration OK

---

./afterburner m

VPP calib. offset: 25
VPP: 4.2 - 5.0V : 5.05
VPP: 9.0V : 9.00
VPP: 12.0V : 12.03
VPP: 14.0V : 13.97
VPP: 16.0V : 16.04

---

Thank you very much!

[AJROMR]

Finally remind you, just in case, that in the file bom.txt where it says:

---

BOM for Afterburner 3.0

• (R1 - 5x) Resistor 4k7 (TH or SMT 0805) 0.25W, 5% or better tolerance - 5x

---

it should read:

BOM for Afterburner 3.1a

• (R1 - 4x) Resistor 4k7 (TH or SMT 0805) 0.25W, 5% or better tolerance - 4x

---

[ole00]

Thanks for the feedback - it's good to know 18k resistor works well. I'm glad it finally works for you.

This makes it easier to test one MT3608 module after another.

Definitely good for testing, but the drawback is the decoupling capacitors are further apart from the booster convertor so they are not as effective filtering the ripples.

It would be better if the board had an extra FB pad just below the MT3608's FB pin.

OK. that can be done - I guess the only reason would be to hide the connection wire? I placed the FB pad away from the MT3608 area to be able to conveniently check the pin and also to make the connection point obvious on photographs for the documentation purposes. The whole confusion about the mt3608 modules (which type works without mods and which type needs a mod) is a bit unfortunate, so I'll probably redesign the area underneath the mt3608 modules and put there extra footprints so that the voltage booster can also be built from discrete parts as an option.

Finally remind you, just in case, that in the file bom.txt where it says...

Thanks, previously there were 5 resistors, one was removed recently. There is a chance I'll put the 5th one back (I'll need to do more tests) - it seems it's not critical to have the 5th one there though. I'll update the BOM in the meantime.

[ole00]

I think this issue/solution might be helpful to others, so I'll keep it opened for others as a reference. I usually do not close issues for that reason, it easier to find answers to the same/similar questions.

[ole00]

One more thing. I just noticed your calibration listing. This particular line is unusual:

*Index for VPP 1650 is 114

Usually, when (during the 's' command) the maximum voltage is set to 16.5V then the last index (for VPP 16.5V) should be 128 (not 114 as you have in your listing). It is not a big issue, just an observation (the code and voltages should still work fine). It would be better to retest the VPP values with multimeter while the 'm' command is running to be absolutely sure the VPP voltages are correct.

EDIT: the reason why 16.5V should be on the last pot tap index (128) is safety of the GAL chip. It ensures that in any possible/unexpected error situation (when the POT tap value is maxed) the voltage does not exceed 16.5V. In your case it looks like there is another 14 taps available (128-114), each increasing the voltage by ~ 0.25V which means the VPP voltage on pot tap index 128 may reach 20V which would fry the GAL chip.