240ms @ 45.83A

GTSZ

Condition (Vac, A)	Spec (mS)	Reading (mS)	Picture
100Vac, 45.83A	2000	276
240Vac, 45.83A	2000	170

@Vision314

Delay is in milliseconds, FYI.

Suzhou tested 100Vac and 240Vac. We only specified at 115Vac. If we are going to include their data, the 100Vac condition is more relevant. (240Vac attached above.)

sounds good, added both

GTSZ

1% ripple = 120mV

ALL PASS

Output Voltage	Picture	Measured Ripple
95% Vout		31.3 mV
100% Vout		46.7 mV
105% Vout		62.2 mV

Load Step

Load Dump

Load Step

Load Step	Picture	Deviation	Recovery Time
0A → 11.5A		237mV	0.2ms
11.5A → 22.9A		133mV	0.2ms
22.9A → 34.4A		140mV	0.2ms
34.4A → 45.83A		147mV	0.2ms

Load Dump

Load Dump	Picture	Deviation	Recovery Time
11.5A → 0A		235mV	Dev. <1%
22.9A → 11.5A		123mV	Dev. <1%
34.4A → 22.9A		147.5mV	0.1ms
45.83A → 34.4A		150mV	0.2ms

@Vision314 See https://github.com/GlobTek-Engineering/GTM965500P/issues/43#issuecomment-4247254592

fixed !

Output Voltage Overshoot

Output Voltage Overshoot

Condition	90 Vac	264 Vac
No Load
Full Load

Screenshot @264Vac, 13.1V output:

Yellow - Output Voltage

Blue - PG signal

Pink - mains voltage

85% load = 37.1A

Part A — Remote Sense Compensation

¹ V_Comp = V_Supply(connected) − V_Supply(disconnected)

Part B - No damage

Part C - No damage

85% load = 37.1A

Part A

Part B - No damage

Part C - No damage

@Vision314

> Part A

>

> 1. 12.608 V

> 2. ..

> 3. 12.86 V

>

The voltage has only increased by 252mV (if judging by 12.86-12.608). Our spec is 500mV (min.). Have you confirmed the length of wire you are using is sufficient to generate at least 500mV drop at the load current?

(You can just apply a 37.1A load (without remote sense attached) and check the difference in voltage at the PCB versus at the load to confirm.)

> * voltage @ load is only about 12.5V, should be 12.6

This might be considered "normal", as when SNS leads are connected, there is still *some* contribution from the sense points on the PCB, so we may not expect perfect regulation. Perhaps we could add a "load regulation" line to the remote sense part of the spec...maybe +/-2%.

@Vision314 Good. I suggest updating the SNS results tables to something like this. This shows the "natural" drop of the cable, and then the amount of voltage the SNS circuit was able to "add back", and the spec.

Req of Ra8, Ra9 and Ra11, Ra12 is 45.3kOhm

Ra8, Ra11 = 44.2kOhm

Ra9, Ra12 = 1.1kOhm

Ra14 = 18kOhm

Output range: 11.4V -> 12.6V

For future reference:

@Vision314 Good, can you make a separate 'result' tag for the two voltage measurements after the change?

And flag the "Changes required" field as "Yes"...although I guess that might be irrelevant if we are parsing and checking for the 'change' word

I think these values are the best way to go

OPP point is within specified range, but PG time for 90Vac typ and min output voltage is too quick. I suspect this is a problem related to the PFC at 90Vac. looking into it.

OPP Point and PG time for new values:

@Vision314

> OPP point is within specified range, but PG time for 90Vac typ and min output voltage is too quick. I suspect this is a problem related to the PFC at 90Vac. looking into it.

PG time >100% load is a non-specified condition, so let's not worry about it. That said, PG time is "0" because the PFC is turning off and PFC does not signal ahead to the LLC controller about that. So, when the PFC turns off, the LLC turns off; there is no warning.

The remaining issue is that OPP @ 90Vac is different than at 240Vac, which indicates a PFC issue as noted. We could also consider this a "don't care" condition since 131 - 135% is within range, but it seems to indicate that PFC is borderline/marginally operating. We should check SNSMAINS or PFC OCP again.

PG time is good at 120Vac anyway.

I'll see what I can find in the PFC.

1. Removed the temperature sensing circuit from the PFC SNSMAINS to see if it was similar to the noise issue we had on the 24V

- No change, so that's not the source of the issue

2. RINGO isn't showing any protection triggers on PFC

3. RINGO was not on the most recent version.

- Fsw min was set to 65k. Reset it to 25k (TEA2376 V9)

- no change in OPP point at 90Vac

4. Will check LLC RINGO next

@Vision314

> Input Voltage Output Voltage OPP Point (A) OPP Point (%) PG Time (ms)

> 12 60 131% 0

> 12 69 151% 10

> 12 71 155% 9.6

12V nominal condition at 120V or 240Vac should be set to 130-135%.

> Removed the temperature sensing circuit from the PFC SNSMAINS to see if it was similar to the noise issue we had on the 24V

Please check whether the PFC voltage starts to drop as the load approaches 130% @ 90Vac. We've seen that PFC voltage droping to ~300V causes the LLC drop out due to undervoltage and/or overcurrent due to lower PFC voltage. (Lower PFC voltage = higher LLC current.)

I would be hesitant to ascertain that SNSMAINS is not the source of the issue, even if removal of the OTP circuit does not change the situation.

PFC voltage is dropping before it hits OPP point at 90Vac, 12V output.

This photo is after I added the SNSMAINS rework from the 24V.

Replaced all the resistors in the instrumentation amplifier on the regulation board just to double check and make sure there's nothing wrong there.

> Noting current values for regulation board. For testing between -5% to +(5% + 500mV) => 11.4V to 13.1V

>

> 11.2 V -> 13.3V

>

> Ra8, Ra11 = 44.2kOhm

> Ra9, Ra12 = 300Ohm

> Ra10, Ra13 = 10kOhm

>

> EDIT: make sure they are 0.1% resistors

https://github.com/GlobTek-Engineering/GTM965500P/issues/66#issue-4147286551

Re-ran OPP test with 11.4 to 13.1 V output voltage range.

OPP point is within specified range, but PG time for 90Vac typ and min output voltage is too quick. I suspect this is a problem related to the PFC at 90Vac. looking into it.

OPP Point and PG time for new values:

@Vision314

> I think these values are the best way to go

>

> COMPONENT OLD VALUE NEW VALUE

> C26 6.8nF 6.8nF

> C27 3.3nF 2.2nF

> Input Voltage Output Voltage OPP Point (A) OPP Point (%) PG Time (ms)

> 90Vac 11.4 68 148% 0

> 12 64 140% 9.5

> 13.1 47 112% 9.5

> 264Vac 11.4 78 170% 9.5

> 12 64 140% 9.5

> 13.1 47 112% 9.5

Looks good. Yes, as a general rule, if you want to give flexibility for tuning (trimming) in the future, and you have two components (either in series or parallel) which do the tuning, and both are using E-preferred values (1, 1.5, 2.2, 3.3, 4.7, 6.8), best to have one of the two be as big as possible (for coarse tuning), and the second is smaller to allow for more granular fine tuning. More granularity with the smaller values because the jump from 1 to 1.5 is small, but the jump from 4.7 to 6.8 is big, for example.

Also, shall we leave this open to indicate that we still want to take a look at the 90Vac / 11.4V PFC drop issue which has not been resolved? Or do we make a separate issue and make some sort of reference to it?

[GTSZ]

100Vac 12V/45.83A	240Vac 12V/45.83A

545 mW

See: https://github.com/GlobTek-Engineering/GTM965500P/issues/24#issuecomment-4255997812

I think we can say this passes because of this test:

https://github.com/GlobTek-Engineering/GTM965500P/issues/66#issuecomment-4165330637

GTSZ

> I think we can say this passes because of this test:

>

> [#66 (comment)](https://github.com/GlobTek-Engineering/GTM965500P/issues/66#issuecomment-4165330637)

Yes, including this "Input Voltage" test is sort of a formality because many other tests will incidentally do the same test.

570mW

See: https://github.com/GlobTek-Engineering/GTM965500P/issues/24#issuecomment-4255997812

239ms @ 22.9A

GTSZ

Condition	Spec (mS)	Reading (mS)	Picture	Result
100Vac, 22.916A	2000	1150		PASS
240Vac, 22.916A	2000	710		PASS

@Vision314

> # GTSZ

> Condition Spec (mS) Reading (mS) Result

> 100Vac, 22.916A 2000 1150 PASS

> 240Vac, 22.916A 2000 710 PASS

Suzhou's delays are much longer than expected. I suspect something is not quite right on Suzhou unit. The turn-on delays should be relatively constant regardless of output voltage. How do we mark this to let them know as an FYI?

I marked as re-test/clarifiation req'd for now.

I think it's better to mark this as has issue, and we can resolve it later

GTSZ

ALL PASS

1% ripple = 240mV

Output Voltage	Picture	Measured Ripple
95% Vout		29.2mV
100% Vout		51.7mV
105% Vout		84.2mV

Load Step

Load Dump

1% = 240mV

Load Step

Load Step	Picture	Deviation	Recovery Time
0A → 5.7A		493mV	Dev. <1%
5.7A → 11.5A		248mV	0.188ms
11.5A → 17.2A		255mV	0.188ms
17.2A → 22.9A		303mV	0.187ms

Load Dump

Load Dump	Picture	Deviation	Recovery Time
0A → 5.7A		220mv	Dev. <1%
5.7A → 11.5A		182mV	Dev. <1%
11.5A → 17.2A		181.7mV	Dev. <1%
17.2A → 22.9A		207.5mV	Dev. <1%

In reviewing the transient response for 12, 24V, and 54V, many times the peak deviation does not ever exceed the 1% recovery threshold, so the recovery time should probably be marked as "N/A" or "Dev. < 1%" for those items. (Cannot quantify how long it takes to recovery if it never deviated outside of our bounds to start with.) Above, all of the load steps exceed 1% deviation (240mV), so first we need to assess whether the peak deviation has exceeded the 5% max or not. (All OK.) Then, for those which have exceeded 1% deviation (but less than 5%), check how long it takes to recover to 1%. Seems like most are in the 200-300us range. Writing <1ms is OK as a pass/fail. For the future, you *could* record the actual recovery time.

No need to re-test, just update those which never deviated outside of 1% to "Dev. < 1%" or similar.

As a note:

I'd probably set the cursors as follows:

Y(1): At peak of measured waveform (measured max deviation)

Y(2): At the 1% recovery threshold

X(1): At t=0 (the moment the load step is applied)

X(2): At the moment the voltage recovers to less than Y(2), the recovery threshold (measured recovery time)

It may make sense (in general, all types of tests), to present data with the specified limits presented in line. This way, you can easily see (at a glance) whether the spec is met or not, no need to scroll up or remember what the spec is.

fixed !

Output Voltage Overshoot

Output Voltage Overshoot

Condition	90 Vac	264 Vac
No Load
Full Load

Previous issues were caused by broken/missing/incorrect TVS diodes for ZD6 and ZD7. Must be 54V zener ("RE" marking) for all voltage models.

85% load = 19.5A

Part A

Part B - No damage

Part C - No damage

Part A

85% load = 19.5A

1. 25.3 V

2. ..

3. 25.67 V

Part B

1. When I connected the SNS- to +load, there was a spark between those two terminals and the supply OVP'd. On startup the output voltage goes to ~40V and OVP's.

Part C

1. ...

The ZD7 TVS diode was removed previously for other rework troubleshooting

When the SNS+ connection touched the negative output terminal, this create some large voltage spike that damaged RJ2

RJ2 Failure Mode: open

Replacing RJ2 with another 2.2Ohm resistor fixed the supply

Solution: don't remove TVS diodes, we need them...

I will finish SNS testing shortly

See: https://github.com/GlobTek-Engineering/GTM965500P/issues/55#issuecomment-4256097729

Req of Ra8, Ra9 and Ra11, Ra12 is 90.7kOhm

Ra8, Ra11 = 88.7kOhm

Ra9, Ra12 = 2kOhm

Ra14 = 16kOhm

Output range: 22.8V -> 25.2V

@Vision314 Good, can you make a separate 'result' tag for the two voltage measurements after the change?

And flag the "Changes required" field as "Yes"...although I guess that might be irrelevant if we are parsing and checking for the 'change' word

added !

Trying the same values as the 12V:

FRANKIE TESTING NOTES:

4/1/26

Noting current values for regulation board. For testing between -5% to +(5% + 500mV) => 22.8 V -> 25.7V

21.4 V -> 26.5V

Ra8, Ra11 = 84.5kOhm

Ra9, Ra12 = 100Ohm

Ra10, Ra13 = 10kOhm

Ra14 = 7.5kOhm

PASS

[GTSZ]

100Vac 24V/22.916A	240Vac 24V/22.916A

GTSZ

https://github.com/GlobTek-Engineering/GTM965500P/issues/68#issuecomment-4172278539

560mW

> [RESULT] 560mW

We could:

1. Change the spec to "500mW typ."...a bit of "specsmanship",

2. Change the spec to "600mW max."

3. Analyze where the 500-600mW is coming from...It does seem a bit high for essentially only the aux supply operating.

500mW (max.) *is* a popular spec among the competition, though I kind of doubt customers actually care...

--

Aux supply has pre-load resistors R71 and R72 to improve its cross regulation performance. About 80mW dissipated in R71 (12^2/ 590*3), and 40mW dissipated in R72 (5^2/590). But we cant' really change the values because they affect the cross-regulation performance.

560 - 120 = 440mW

Startup resistors R59 and R60 should only account for maybe 10mW. Bleeder resistors R1 and R2 about the same.

So where's all this remaining power going...hmm..

We saw those big gulps of 4 or 5W on the Yokogawa, right? That's probably it. That might be indicative of LLC or PFC controller trying to stay afloat via SUPHV startup resistors (which is very lossy, but not an issue because it's only supposed to activate once at startup.)

I think we should try to pin-point who is demanding power when we see the big 4-5W (?) gulps. Let's look at it together. We'll take a look at SUPIC. If it does not seem easily solvable, we'll punt on it an just adjust the spec I think.

277ms @ 10.2A

> [RESULT] 227ms @ 10.2A

Since the delay is considerably less than 2 seconds, there's no concern. Turn on delay is the time between AC turn on and when the output reaches 100% * V_out. I realize I did not explicitly write time to 100%, and DC turn-on could be interptetted as the moment the DC starts coming up. For rise-time, which is not this test, measured as the time it takes to rise from V_out * 10% to V_out * 90%.

Here, the second cursor should have been about 50ms farther out, when output voltage reaches 100%.

Suggest re-testing this one, or adding a note to the screenshot indicating cursor does not match up with actual delay, and then estimate the turn on delay by eye.

sounds good, made a note of the estimated turn on delay in the screenshot and changed the result to +50ms, about right...

1% ripple = 540mV

ALL PASS

Output Voltage	Picture	Measured Ripple
95% Vout		76.7mV
100% Vout		103mV
105% Vout		163mV

@Vision314

Good. For the future, I would include all spurious parts of the waveform as part determining the ripple. I know 95% of the waveform falls within the bounds you show, but it's still a bit subjective.

We are very far away from the limit here so it's not really an issue.

1% = 540mV

5% = 2.7V

Load Step

Load Dump

Load Step

Load Step	Picture	Deviation	Recovery Time
0A → 2.5A		400mV	Dev. <1%
2.5A → 5.1A		183mV	Dev. <1%
5.1A → 7.65A		223mV	Dev. <1%
7.65A → 10.2A		210mV	Dev. <1%

Load Dump

Load Dump	Picture	Deviation	Recovery Time
0A → 2.5A		280mV	Dev. <1%
2.5A → 5.1A		200mV	Dev. <1%
5.1A → 7.65A		217mV	Dev. <1%
7.65A → 10.2A		237mV	Dev. <1%

See: https://github.com/GlobTek-Engineering/GTM965500P/issues/43#issuecomment-4247254592

fixed !

Output Voltage Overshoot

Output Voltage Overshoot

Condition	90 Vac	264 Vac
No Load
Full Load

@ max output voltage @ 50% load, the Hold Up time is a little bit slow. but maybe we don't care....

85% load = 8.67A

Part A — Remote Sense Compensation

¹ V_Comp = V_Supply(connected) − V_Supply(disconnected)

Part B - No damage

Part C - No damage

85% load = 8.67A

Part A

Part B - No damage

Part C - No damage

> [RESULT] 85% load = 8.67A

>

> ### Part A

> Step # Voltage @ Supply Voltage @ Load

> 1 56.83 V 56.12 V

> 3 57.3 V 56.47 V

57.3 - 56.83V is 0.47V, which means the remote sense circuit added 470mV to the output voltage, which is still short of 500mV. Question is whether the wire resistance is still too low or if it is indeed limited to ~470mV of added compensation.

See: https://github.com/GlobTek-Engineering/GTM965500P/issues/55#issuecomment-4256097729

@Vision314

> > [RESULT] 85% load = 8.67A

> > ### Part A

> > Step # Voltage @ Supply Voltage @ Load

> > 1 56.83 V 56.12 V

> > 3 57.3 V 56.47 V

>

> 57.3 - 56.83V is 0.47V, which means the remote sense circuit added 470mV to the output voltage, which is still short of 500mV. Question is whether the wire resistance is still too low or if it is indeed limited to ~470mV of added compensation.

Actually, we have all the information we need. I apparently just got lost interpreting the data the other day.

The cable/resistor is producing 0.71V of natural drop, but the SNS circuit is only compensating/adding an addition 470mV. Can you double check that Da1 and Da2 (on the regulation board) are the same type as on the 12V and 24V? We didn't happen to swap those out before, right?

It could just be that we are inherently limited at 57.3V due to LLC limitation @ 85% load? Try repeating this test at ~95% load. If the voltage at the supply does not reach 57.3V as it did at 85% load, maybe the LLC has run out of gain and so its inherently limited.

Maybe an interesting point for units running below the resonant frequency (i.e. not a multiple of 12V). They are already operating under boost gain (up the LLC gain curve, to the left of the resonant frequency), so they might not have much boost left.

Ra8, Ra11 = 200kOhm

Ra9, Ra12 = 4kOhm

Ra10, Ra13 = 10kOhm

Ra14 = 18kOhm

Output range: 51.3V -> 56.7

@Vision314 Good, can you make a separate 'result' tag for the two voltage measurements after the change?

And flag the "Changes required" field as "Yes"...although I guess that might be irrelevant if we are parsing and checking for the 'change' word

added the separate result and change tags

i think it'll be more clear to just have "changes" and since they are already specified for a voltage model or a model level change, it will make sense when we look back at it in the future.

https://github.com/GlobTek-Engineering/GTM965500P/issues/67#issuecomment-4172429455

GTSZ

12V

Spec (A)	Reading (A)	Result	Picture
6.5	6.128	PASS

24V

Spec (A)	Reading (A)	Result	Picture
6.5	6.056	PASS

[GTSZ]

12V

Voltage	Spec (A)	Reading (A)	Result	Picture
115Vac	35	14.4A	PASS
240Vac	70	44A	PASS

24V

Voltage	Spec (A)	Reading (A)	Result	Picture
115Vac	35	18A	PASS
240Vac	70	38.4A	PASS

GTSZ

24V, 115Vac, Average Efficiency = 92.102%

24V, 230Vac, Average Efficiency = 94.247%

12V, 115Vac, Average Efficiency = 91.907%

12V, 230Vac, Average Efficiency = 93.553%

YELLOW - PG Signal

PINK - Output Voltage

12V

Failure Mode	PICTURE	PG Time
OPP		9.5ms
AC Power Failure		11.8ms

OPP Load = 66A (144%)

75% load = 34.4 A

24V

Failure Mode	PICTURE	PG Time
OPP		9.7ms
AC Power Failure		12.2ms

OPP Load = 32A (139%)

75% load = 17.2 A

54V

Failure Mode	PICTURE	PG Time
OPP		9.7ms
AC Power Failure		12.4ms

OPP Load = 13A (128%)

75% load = 7.6 A