800w HMI Ballast repair help with determining if transformer is bad SOLVED

[jacamo]

Hello I am hoping someone can help me with this repair and determine if T1 is indeed bad which I think it is.

I have attached a schematic and when I got the unit there was some pretty obvious signs of damage to R23 and when I went to remove U1 from its socket the IC just fell apart and I could tell it suffered a catastrophic failure. Testing around that area reveled that Q3, D17 and R37 were all testing bad.  After replacing these components and not finding anything else that was testing weird I tried to power the unit up with no results.

The schematic in my opinion is drawn in a very confusing way and I had to redraw a few sections to try and understand what they are doing which helped. I still am a little confused about somethings but think I understand the general operation.

Also SW1 is always closed in normal operation. On the actual unit this is a voltage selection switch and when set to 115V it is always closed. There is also a switch on the hot side that is not pictured in the schematic I have attached which is used to turn on the unit.

The main issue I can tell with it not working is I am not getting a steady 15VDC power rail.

When turning on the switch I am getting ~170V at the point listed as 325V and this is I am assuming because I don't get K1 to close because I don't have that 15VDC rail.

U1 is being powered by the unregulated bulk DC which is clamped with a 20V Zener (D14) so it shouldn't have any problem turning on. However the voltage is present and stable 20V when U1 is not installed. When I install U1 and check the supply rail again I am getting a very bouncy voltage between ~10-15VDC and when looking at it on my scope it almost resembles a ramp wave form. I think this is the ICs attempt to keep turning it self on but something is preventing it from doing so. Which I think is a short in the primary of T1

I have a good working unit that I measured the primary of T1 and was getting around 6.5Ohms. When testing the unit that is not working I get 0.3Ohms. This leads me to believe that T1 has a short in the primary but wanted to get someone else opinion.

The only other thing I can think of it being is when I ordered parts I went off the schematic and got the UC3844 but what is in the actual unit is a UC2844. From what I can tell from the data sheet is they are basically the same part just one has a higher temp rating and tighter tolerances.  I don't think that is the issue though just thought it was worth mentioning. 

Honestly hoping it is not the transformer as that's one of the hardest things to find replacement for. But if it is I would like to confirm that its bad before I begin searching for a replacement. If anyone has any insight as to what could be causing this problem and if it is T1, any experience with finding a replacement transformer like the one in this unit. It seems to be a custom wound one as I couldn't find any info on the part number and the sticker on it has "Current Controls" which a quick google search said they make custom transformers and the like.

Thanks for the help

[jacamo]

Well since I never got any replies about this I went ahead and figured it out on my own.


Testing T1 out of circuit a few ways reveled it was indeed shorted on the primary. I ended up taking it apart and rewinding the transformer. The circuit is now working as it should with one exception. The output voltage is now 16.3VDC instead of 15VDC.

Now I really don't think this will affect the operation of the rest of the circuit as all the other ICs powered from that rail have voltage ratings well above 16VDC.

I am going through the data sheet for U1 which is a UC3844 Data sheet linked here
https://www.ti.com/lit/ds/symlink/uc2845.pdf?ts=1673642593950&ref_url=https%253A%252F%252Fwww.google.com%252F

I am assuming that I can change some values of resistors in the feedback path to lower the voltage to get 15VDC but can't seem to figure how how to calculate the right values needed.

As stated in the data sheet "During normal operating mode, the IC can be used in peak current mode or voltage mode control. When the
converter is operating in peak current mode, the controller regulates the converter's peak current and duty cycle.
When the IC is used in voltage mode control, the controller regulates the power converter's duty cycle. The
regulation of the system's peak current and duty cycle can be achieved with the use of the integrated error
amplifier and external feedback circuitry." 


The application they use in the data sheet is a bit different than what is in the circuit I am repairing so It might be confusing me.

If anyone has any experience with these ICs I would love a little bit of guidance.

Thanks

[timeandfrequency]

Hi jacamo,

My answer comes a bit late as you already successfully repaired your unit.

And luckily you've got the schematics, which, of course, is of great help.

As you are located in the USA, I assume that you use 120VAC line voltage to power the gear.
HMI/HQI lamps electronic ballasts are not that easy to understand because, functionally speaking, it is an AC current source plus a high-voltage igniter. The arc lamps need a significant warmup time (up to 5 minutes) which has to be very carefully monitored. Once at maximum brightness, the arc voltage for many usual bulbs is about 95VAC.

So it's not easy stuff and this unit doesn't make an exception : the 'power section' consists of a mains filter, an inrush current limiter (K2), then a voltage doubler (K1), followed by two parallel buck regulators (Q4, Q5) and finally a full H bridge forged around four IGBTs.

And you also made the correct assumption : U1 plus T1 form an auxiliary power supply and if not functional, the unit won't work correctly.
Rewinding T1 which is probably a ferrite transformer, is definitery a significant achievement. Quite all of them are custom made : so even if you ask the manufacturer of this electronic ballast, you might not be able to buy a replacement part. At best, he will be able to provide the complete 'power board'.

And yes, for this kind of gear, you can replace the UC2844 with an UC3844.

Powering the UC3844
Have a look at Figure 8-7 for a ballpark view on how it works.
On your schematic, the UC3844 starts when the voltage across C22 >= 16V : the current comes from the 325V rail via R10.
Once running, the UC3842 receives its power supply from the 5-6 winding of T1, via D8 and D7.

Output voltage regulation and calculation
The UC3844 is a current-mode PWM controller which also has a voltage feedback loop, as explained at § 'Pin 2  Error amplifier inverting input'.   
The internal comparison threshold is 2,5VDC (= VREF/2).
In your circuit, the output voltage probing is done by R24 and R21. I went through several datasheets and appnotes talking about those PWM controllers. It's a bit shabby, but it seems that the formula for calculating the output voltage is not provided anywhere. You have to analyse the application examples and recalculate the formula. Here it is, with the actual parts references used in your schematic :
Output voltage = [1 + (R24/R21)] * 2,5 = [1 + (24,9 / 4,99)] * 2,5 = 5,99 * 2,5 = 14,97V which is about 15VDC as expected.

As you measured 16.3VDC, and even if it seems not to harm the other sections of the electronic ballast, there might still be a small issue pending : you should actually be very close to 15 VDC.
As it's not that bad, you may leave it as is, or if you want to dig further, go on reading below.

To go further
If you could provide the schematics of the control(s) board(s) (connected to P1, P3 and P4) and the igniter board (connected to P2), this would help to validate if 16.3VDC as AUX power value is acceptable or not.
16.3 VDC is most of the time a still acceptable value for a 15VDC power supply, but it should be rock on 15VDC : as it is not the case, this might hide some ringing or even an U1 hiccup mode or IC failure.

As there's a regulation loop inside U1, the fact that you perhaps did not exactly wind the same amount of turns for T1 should not influence the actual output voltage : the loop shall be able to compensate. You are right, the UC3844 is really not easy to understand because of its combined current and voltage steering which shall improve its loop stability. But it would be overkill to validate the loop stability of your unit : it requires very specific and expensive test equipement.


The AUX power supply is not insulated from the mains
Even if U1/T1 looks like a flyback power supply, the 15V rail is not isolated from the mains because the T1 secondary is tied to the ground. And the latter is derived from the mains via BR1 pin 4.
This means that when this gear is powered, you should only use a multimeter to probe inside. To avoid any electrical shock, an isolated oscilloscope or a high voltage differential probe would allow to keep you on the safe side, while looking for an in-deep view of what's going on.
 
R19/C19 sets the oscillating frequency to 78.2 kHz (calculated value). But for the UC3844, this value has to be halved, so it is 39.09 KHz.

Other checks
Now that T1 is repaired, check again U1s power supply (pin VI = red probe, pin GND = black probe). It should be between 11.5VDC (because below, the UC3844 detects instantaneously an 'UVLO condition' and restarts/reboots) and 20VDC, which is the D14 zener voltage. You might want to replace C11, C12, C17 and C22 and see if it helps.

Then check VREF and it should be 5VDC
Then check VFB  and it should be 2.5VDC

With the unit turned off, also check the resistance value of R21 and R24. When measured 'off-circuit' (= one leg lifted), they should match within 1% of their indicated values.
Also check D7 and D8.

Please report the values you measure.

[jacamo]

Hello and thank you so much for the reply. I have read it a couple of times already but its quite a lot of info to digest at once. I am not at my shop either so I won't be able to measure the requested things but I will as soon as I can.

I just wanted to thank you as this helps me out a lot with understanding and confirming what I had figured out. I do have the full set of schematics and I will attach the control board. I have attached 2 different versions of the control board and the one that is actually in my unit is the one labeled ContolCircuit2 but I though the other one was drawn a bit better with the connectors to match up with the power board schematic. After looking at it again the 15V rail might have to be dead on 15V for timing reasons.

You are correct when it comes to finding a replacement for that transformer. I called the manufacture of the ballast and they don't have any replacements or even old "donor" boards they could sell me. This is a pretty old unit and I didn't really think they would have it but was worth a shot. I decided to try and rewind the it since I didn't really have any other options. I posted a photo of in the middle of unwinding the secondary which I had to unwind both of them because they were on top of the primary. The last photo is the completed transformer rewound.

I have not yet tried to hook up a lamp head and fire it up under full load until I could figure out how critical that 15VDC rail is. I will preform the tests and measurements you have requested and will have a bit more info. Especially with your feedback I am excited to learn more about the in-depth operation of these ballasts. I work on them quite a bit.

Once again I am very appreciative that you took the time to respond with so much detail. Thank you and will get back to you as soon as I can.

[jacamo]

The AUX power supply is not insulated from the mains
Even if U1/T1 looks like a flyback power supply, the 15V rail is not isolated from the mains because the T1 secondary is tied to the ground. And the latter is derived from the mains via BR1 pin 4.
This means that when this gear is powered, you should only use a multimeter to probe inside. To avoid any electrical shock, an isolated oscilloscope or a high voltage differential probe would allow to keep you on the safe side, while looking for an in-deep view of what's going on.
 
R19/C19 sets the oscillating frequency to 78.2 kHz (calculated value). But for the UC3844, this value has to be halved, so it is 39.09 KHz.

Other checks
Now that T1 is repaired, check again U1s power supply (pin VI = red probe, pin GND = black probe). It should be between 11.5VDC (because below, the UC3844 detects instantaneously an 'UVLO condition' and restarts/reboots) and 20VDC, which is the D14 zener voltage. You might want to replace C11, C12, C17 and C22 and see if it helps.

Then check VREF and it should be 5VDC
Then check VFB  and it should be 2.5VDC

With the unit turned off, also check the resistance value of R21 and R24. When measured 'off-circuit' (= one leg lifted), they should match within 1% of their indicated values.
Also check D7 and D8.

Please report the values you measure.

Hello again,

I am using a Isolation transformer and a HV Diff probe when looking at anything that is ground referenced to the mains.

I had made the same frequency calculation yesterday based on the component values and the formula from the data sheet and am got the same result at 78,181Hz but I don't quite understand by what you mean when you say it its half for the the UC3844? Could you explain what you mean or refer to where in the data sheet it explains that?

As per your request I went ahead and check the Voltages

U1s Suppy Voltage is 16.23 check on the IC

VREF  5.029VDC
VFB 2.500VDC

Resistors were well within 1%

I then realized that the rail I was checking was the voltage for U1 and not the 15V rail..... Checking it at the right place it is 15.39VDC so I think I can call this part of the circuit working. 

I still need to test it under load but so far so good.

[timeandfrequency]

Hi jacamo,

Thanks for your kind reply and the additional documentation, and also for the latest update.
Just take as much time as you need to probe the unit, there's no schedule or deadline on my side.

Wow ! 15.39VDC for the AUX power is really good news ! Leave it as it is.
With U1s supply voltage at 16.23VDC, we are also really confortable (= just above 16VDC which is its turn-on voltage; turn-off voltage is 11.5VDC, worst case).
The VREF & VFB values you measured at U1 are exactely what we expect, so the PWM controller actually works fine. Well done !


Using an isolation transformer and a differential probe is really a good practice when probing inside those units.
Looking at the pictures, I must also admit that rewinding that tiny transformer was really a demanding task. Nice work !


More about K1
I also have to be more specific about K1. Of course, K1 is a relay and not actually a voltage doubler as I misstated in my previous post. In fact, K1 is the relay that - when energized - activates the voltage doubler feature.


Why do we need a voltage doubler ?
325VDC is a very common DC rail voltage value, so the answer has something to do with geographical location. In Europe, the mains is 220-240VAC. E.g. in France we have 230VAC. The peak value of a sinewave measured at 230VACrms is exactly 325Vpeak. And thanks to the big caps C5, C6, C8 & C9 this becomes 325VDC.
So the voltage doubler is just here to always have a 325VDC rail, should the unit be used in the States (120VAC) or in Europe (220-240VAC).


Is it important for the AUX power to be spot on the value of 15VDC ?
To summarize :
I just had a quick look at the 'ControlCircuit2.jpg' schematic.
The 15VDC rail is not just used for power supply purposes as we would normally expect. Its voltage value determines part of the behaviour of the arc lamp power supply.
Not being rock on its nominal value will probably slightly shift the lamps' working conditions and influence the startup sequence. But I cannot say that it's a no-go. With 16.3VDC, we are just a few percent off, and when looking at the schematics it's still acceptable.
-> Now that you actually measured 15.39VDC, well, your issue has just vanished.


AUX power is also somehow used as reference voltage on the 'control board'
It is not really the best practice to use a power rail simultaneously as a power supply and as a voltage reference (even if there's also a second more precise voltage reference '+VR' on the 'control board'). But if thoroughly tested and if no erratic behaviour shows up, we have to admit that this might be scored as an acceptable design, because it actually works.


Unknown connectors
On the 'control board' schematic, I noticed other connector labels from J6.1 to J6.6.
There's even a J7 connector below U1.
Do you have any information/schematics about the boards tied to J6 & J7 ?


Clean or noisy AUX power ?
Even if its voltage value is a bit off specs, what is important is that the AUX Power line is clean with relatively low ripple. I would say that 100 mVpeak_to_peak should not be exceeded when measured while the 'control board' is disconnected.


Alternative to the existing AUX power supply
Disabling the legacy AUX power supply around U1 and T1, and using an industrial fully enclosed 15VDC switching power supply with some additional filtering at the output might be an acceptable replacement part, should you again have to cope with such an issue. You will also be able to connect it to the 325VDC power rail, because the selected model accepts up to 375VDC at the input.


Internal oscillator frequency of the UC3844 chip
I could find that information in its datasheet, about the RT/CT pin : the frequency of the UCx844 and UCx845 is equal to half of the fOSC frequency at up to 50% duty cycle.
I did not verify by myself, but as it is written, this would mean that the actual PWM frequency at pin 6 'OUTPUT' is only 39kHz. Doing it that way avoids going above 50% duty-cycle.


A few interesting links
HMI's and Electronic Ballasts - Frequently Asked Questions
Attempt to built inverter “ballast” for 4kW HMI lamp
This company seems to provide repair service for electronic ballasts.




[UPDATE 1] : Voltage doubler feature is activated by K1 relay, not K2, as initially written.

[jacamo]

Hello,

I had been waiting for parts and now that they have arrived and installed I wanted to give an update on the ballast and had a few questions.


K1 and the Voltage doubler

I had read that one of the reasons they use a doubler is to help with the PFC. I understand the whole reasoning behind the different voltages based on where you are located in the world. but was wondering how much it would help the PFC?

Aux power and the reference voltage

I can see why using a voltage rail as a reference could be a bad think but like you said it works. I was also wondering about how the UC3844 uses the secondary 5-6 to power it self after startup? Is this something that is done often?


Unknown connectors

J7 is not used on the ballast I have. It says that "J7 and R28 not used on all models" 
J6 is however an unpopulated connector that I am assuming was used for probing and testing.

Clean or noisy AUX power

You can see in the screen shot of the scope photo I attached that I am getting really bad ringing or something on that 15V rail. This is the same with the control board still connected or not. I am not quite sure what to make of this but considering that this is used as a reference on the control board I don't think it is acceptable.

Alternative to the existing AUX power supply

This is a good idea and might be the option I take to get this totally back to factory specs.


Internal oscillator frequency of the UC3844 chip

I understand what you are saying now.


Thank you for the links and again the very detailed reply.

Hopefully I can figure out what is going one on that 15V rail.

[timeandfrequency]

Hi jacamo,

Here are some answers to your questions.

K1 & PFC
Well, the schematics are not that straightforward to understand, and maybe some other forum members would also like to provide some advice.

There's no explicit PFC in the schematics. Now, do Q4 & Q5 help to smoothen the mains current ? I would rather say no : those are rather used for dimming the lamp.
Does the K1 activated voltage doubler help to improve the power factor ? Again, I don't think so. I've never heard that a voltage doubler running at mains frequency really improves the power factor.
Furthermore, to get a flicker-free beam, it requires the AC lamp current to be asynchronous to the mains frequency (which usually serves as video refresh rate) : this probably worsens the power factor.

To remove any doubt : an efficient PFC sets the power factor to 0.98-0.99. If the 'power board' really acts as a PFC, you should get a pure sinewave when measuring the mains current at TB1, and it has to be in phase with the mains voltage.


UC3844 power-up sequence
What we have is a really vanilla scheme used in many SMPS controllers designed for the flyback topology. A startup resistor (R10) draws some current to wakeup the controller. The latter immediately starts to switch, which leads to a current flowing through the flyback transformer T1 (therefore storing  some energy) whose primary auxiliary winding (5-6 in your case) powers the controller during normal operation.


Noisy 15VDC rail
You are right, the 15Vpeak_to_peak noise on the 15V rail is not nice at all.
First, you need to validate its amplitude.
Much care must be taken when probing : a 10 cm ground lead attached to the probe is a significant antenna that catches every radiated EM pulse coming from unshielded magnetic parts.
It is wise to always use a ground spring to probe faithfully in switching power supplies. A DIY low impedance probe can also be useful (using a mains isolation transformer to power the unit under test is compulsory in that case).

The pictured ringing pulse is of rather high frequency (2.1 MHz). That means :
- it won't bother too much slow acting electronics, like what can be found on the 'control board' (not more than tens of kHz). But a harmful effect should not be undermined.
- if it's a conducted noise, adding a few parts should improve the situation.

While setting the scope input channel to DC mode, can you mesure the repetition frequency of the pulse (e.g. each time time Q3's switched on). Is it 39 kHz ?
The ringing might also be worsend by a light load condition, especially when the 'control board' is not connected.
Try to load the 15 VDC power rail with a 100 Ohms 3 W resistor (to draw 150 mA), an check again.


Clean up the noise on the 15VDC rail
Unfortunately, there's no guaranteed recipe. Some of the EMI and SPS gurus will certainly be able to spot on the issue at first sight.
I definitely cannot pretend to play in that league, so let's give it a try in 'trial and error mode', which is the best I can offer.

You might apply the different changes below simultaneously, even if some nasty interaction cannot be excluded.

a) At first, put one or two ferrite beads on the gate leg of Q3, as close a possible to it's plastic housing.

b) The need to redo the winding of T1 probably slightly changed its inductance and parasitic capacitance. The snubber C25 - R36 might be tweaked to match with the new characteristics.

c) You might also want to add a calculated RCD snubber

d) Add a 15V AC transil in parallel to C13.

e) Add a 100 pF capacitor in parallel to C13

f) Add a leaded ferrite bead in series with D9 (cathode side). Or convert the flybacks output filter to a LCLC filter, by duplicating the ferrite bead and also C15 // C13 // 100pF. The actual component values are not really important.

g) Put a small capacitor (10 pF) in parallel to D9.

h) Try to shield T1 with a thin iron sheet (0.2-0.4 mm), connected to GND (e.g. at pin 5 of T1).

[jacamo]

Hello,



Noisy 15VDC rail

I went ahead a probed the 15V rail again with the ground spring instead of the ground lead and it makes a huge difference. I am now only getting about 20-30mVpp.


Also the second screenshot is of the switching frequency of Q3 which looks to be about 42Khz so not too far off the expected 39Khz.


I went ahead and tested it under full load by firing a light and it seems to working great!


I will reassemble everything and run it for a few hours and see how it goes.


Thank you so much for the help!

[jacamo]

Hello again,

Just wanted to give another updated after running it for about an hour. The reason I cut the test a bit short was because of how hot D9 is getting.

The datasheet below is for the diode and from what I can see the max junction temp is 175C. I was measuring about 132F which is about 55C so thinking about it now I am well below the max temp.
https://www.mouser.com/datasheet/2/427/byv26-1768002.pdf

I was thinking about replacing it just to see if it would make any difference but I have my doubts. The other diodes next to it, D8 and D7 are not getting hot at all but I believe this is because D9 is the one with the most of a load on it.


Also another thing was when measuring the input I am reading exactly 900W of power being used by the ballast after the striking and start up sequence had finished and settled. This is rated an 800W ballast and am thinking that this is due to the losses in the ballast. It also has a pretty dismal power factor of 0.66. I am pretty sure if I measured the power going from the ballast to the head I would get about 800W.

I also went ahead and replaced all the electrolytic caps as a matter of preventive maintenance as this ballast is pretty old and those components are getting close to end of life.

Overall I am feeling pretty good about this repair and the longevity of its use. These units get worked pretty hard out in the field and the situations can be pretty demanding at times. If there is anything else I could do to verify that that this ballast is working at 100 % please feel free to give me suggestions.

Thanks again for your help and feedback. I really appreciate it!

[timeandfrequency]

Hi jacamo,

Wow ! Let there be light.

Only 20mVpeak_to_peak of switching noise for an SMPS where you winded the flyback transfomer manually is a huge achievement.

The signals catched on your scope are really clean (worst case noise is about 30mVpeak_to_peak), showing that everything works fine : for such a basic flyback SMPS, you cannot ask for more.


D9 measured temperature
Honestly, 55°C seems still acceptable, but - it's a bit the same story than for the noise pulse - we need to validate the value. The IR image clearly shows that D9 is hotter than other parts, but the accuracy of the measurement cannot be guaranteed, because measuring a precise temperature with an IR cam depends on the emissitivity of the surface being mesured. And the IR camera setup has to adjusted to that same emissitivity value.

The emissitivity value for different types of glass varies from 0.7 to 0.97, which might be a bit far from the default value set in the IR cam, which is usually 0.95.
Furthermore, the actual emissivity of the part might be botched by the tin on the legs (its emissitivity value is only 0.06).
TELEDYNE-FLIR tells us that if the emissivity of the target you are trying to measure is below 0.5, you are unlikely to be able to get an accurate temperature measurement. The article also focuses on training and certification. IMO, it's worth considering when using such devices for professional matters, because IR cams don't always tell the truth. Especially when it comes to thermal imaging for buildings and renewable energy applications, where these devices are widely used, most analysis and conclusions are wrong due to a lack of training.

The easy way to get accurate temperature measurment, once you spotted which part gets hot, is to use a thermocouple or Platinum sensor instrument and some thermal grease for best possible heat transfer between the part and the sensor.
You may check out those instruments, made by a renowned brand, which are deadly accurate (± 0.2°C), feature an auto-verification function at startup, an the displayed unit can be selected between °C or °F :
- Hanna HI98501
- Hanna HI98509
- Thermal grease

The sole limitation of those devices is the size of  their sensor which is a bit too bulky: when measuring the temperature of a small part, you also need a small sensor, to catch as less heat as possible. Thermocouples devices are sometimes preferred for that purpose, because of their really tiny sensor. Accuracy is about ± 1°C, but iit's sufficient for temperature verification on electronic parts. Caution is advised when measuring the temperature of high voltage parts: thermocouple sensors are not insulated.


D7 and D8 are not getting hot at all
That's not very surprising and your assumption is correct. The current that goes through D7 and D8 is the UC3844 operating current  (17mA worst case) plus the average gate current of Q3 (Qg = 38nC, fsc = 42kHz) which is 1,6 mA. Total current is 18,6 mA, worst case. Dissipation in each diode is therefore about 18 mW.
The current that flows through D9 is significanly higher because it's the supply current of the whole 'control board' plus the average gate current of Q6, Q7, Q8  & Q9 and also the current through T2:1 that has to drive two additional mosfets (Q4 & Q5), and their associated active clamp circuit, plus - finally - the two K1 & K2 relays.
The calculation of the current through D9 is really complex but a ballpark could be between 200 & 600 mA.
Let's calculate the dissipation of D9 @400 mA : it's about 600 mW. The exact value is difficult to calculate because Vf span is 0.35 to 2.5 V !


Alternative for D9 (BYV26C)
if you are not confident in keeping the legacy D9, you may replace it with a STPR1040CTW.
It's a bit bulky but there's not much choice in ultrafast recovery diodes.
The 600V breakdown voltage is not really required here.


Measured power factor of 0.66
The value you mesured does not really suprise me. It actually confirms my assumptions provided in my previous post : there's no PFC feature in this gear.
Only an SMPS with PFC will feature a power factor close to 1.
An arc lamp is a very demanding load to drive as it requires an AC current source. In addition, the long wake-up time of the lamp must be mastered by avoiding injecting too much power into it.


Replacement of old parts
Yes it's wise to replace all of the electrolytic caps, without forgetting the small ones (= small shape), because their life span is lower than the big cans.

[jacamo]

Hello again,

D9 Temp

I went ahead and measured the temp of D9 with a K type thermocouple and my Fluke 87V. As you suggested I went ahead and slathered it up with some thermal grease and the highest temp I am getting is pretty much the same as the thermal camera. 56C.
There is an enclosure that it goes inside of but it is pretty well vented on all sides so I think even with the enclosure it wouldn't get but a few degrees hotter.

The thermal camera I am using is pretty new and is linked below, Mikeselectricstuff does a teardown of it on youtube if your interested. Well worth the money IMO.

https://www.xinfrared.com/products/infiray_p2_pro_thermal_camera


Alternative for D9

Im not sure if you noticed but the part you suggested is in a TO-220 package and D9 is in a SOD-57 so I don't think that would work.
Now that I have double checked the temp and ran the ballast for around 4 hours I am feeling like this can live the way it is.


All the Capacitors

Yep I got them all, There are a few tantalum caps on the control board that I was considering replacing as I have heard about them failing before. What do you think?

[timeandfrequency]

Hello jacamo,

Thanks for the update.

D9 actual temperature
Nice that you could check with a thermocouple. 56°C is fine.

D9 package
I know that the part I suggested does not have the same package, because it's a 10A diode. The TO220 package dissipates heat much better than a SOD-57 package. Functionally, I choose a similar ultra-fast recovery diode, but that part will be cooler than the legacy diode.

IR cam
Nice little stuff this InfiRay P2 Pro. I have been using IR cameras for a good ten years. An invaluable tool for repair and quality control.

Tantalum capacitors
It is wise to replace them too, because their failure mode often leads to a short circuit. As there's no fuse on the AUX power rail, avoiding shorts is good practice. You may replace them by low ESR Electrolytic capacitors from first tier manufacturers like Panasonic, Cornell-Dubillier, Nichicon or Rubycon.