I am trying to convert a dozen non dimming spotlight tracks from tungsten halogen to LED. Each has 3 x 35W bulbs but it will take up to 150W in total. They have a 230VAC to 12VAC inverter transformer at the end of each track and one has failed.
In UK I bought some Integral ILMR16NC033-3 5W warm white MR16 LEDs from Toolstation and similar LAP ones from Screwfix so I have 15W of LED on each track. I can’t find a transformer that will work with them. All 4 LED transformers from 3 suppliers (claimed 16 to 24W output) that I have tried cause flashing of the LEDs indicating current overload. In all cases if I turn it on with less bulbs and then insert the others later it will not flash. So it’s not the steady load that is causing an overload. Or putting the switch in the transformer secondary rather than the primary might lead to a successful start-up.
I think the problem is that these bulbs must have a constant current driver chip such as https://www.st.com/resource/en/datasheet/stcs2a.pdf These have small capacitors in circuit which draw a spike of current when turned on. This inrush current spike may be greater than the transformer’s overload circuit allows, even though it is only 10millisconds long as measured with a scope. See attached traces. These were taken using a LED transformer across 0.1ohm in series with the bulbs so 100mV = 1amp. The second trace is with switching on the secondary side.
Or more seriously, as the transformer voltage rises upon switch-on the LED driver may try to consume more than it’s 12V current. This is because a 15W LED load may consume 1.3A at 12V but, as the transformer output rises through say 6V, it tries to consume 2.6A to deliver 15W, if the bulb’s IC has started up faster than the transformer has. The designer needs more care in designing the transformer so the rise times and time constants are correct. Any overload detector needs a delay circuit of at least 10ms to stop it overreacting to brief high currents. Some LED bulbs may have a soft start feature in them, which may eliminate the inrush, but my oscilloscope indicates that these are not soft start.
One transformer manufacturer stated that the problem was Power Factor in that these bulbs do not have a unity PF. Although a LED transformer has a DC output it is far from smoothed with harmonics from 50Hz to >100kHz. So he said I should aim to keep draw to say 50% of the rating. Interestingly though the one 24W rated transformer I tried was even worse than their 18W one. I noted though that the internal designs were quite different so it’s not just a slightly uprated variant of the same design. I don’t have room for too high a rating transformer.
Another issue is rating. Some transformers rated at say 18W may in reality consume 18W but only deliver 80 - 85% of that, which means that a 15W load is on the limit of what it should deliver. However one manufacturer’s claimed in literature and label that it’s a 18W output transformer but in an email implies 80% of that.
I am tempted to go back to the original tungsten bulb transformers or modern equivalent, if it has failed, as the bulbs do work well with these as they accept an AC input. But my Picoscope software records the RMS voltage as around 6V on such a low wattage load even though the bulbs' brightness ems to be OK. The load is below the minimum these transformer are rated at. A compromise is to fit one tungsten and 2 LEDs to keep it closer to the minimum rating and then 12V is output.
The waveform is horrendous with variable length bursts of many kHz every 1ms. It amazes me that a https://www.st.com/resource/en/datasheet/stcs2a.pdf or similar can cope with such a variable input. Whether it will be reliable long term remains to be seen.
My conclusion is that the manufacturers have designed their transformers to work with the kind of LEDs with a dropper resistor and no IC as these will accept any sort of pulsed input and not have inrush issues. But it’s impossible to tell what sort of LED one has bought by looking at it and the mainstream sellers like Screwfix don’t sell them.
If space allowed I’d just put a 230 to 12VAC standard non electronic transformer to power several tracks at once and ditch all these dodgy inverters.
Any ideas please?
In UK I bought some Integral ILMR16NC033-3 5W warm white MR16 LEDs from Toolstation and similar LAP ones from Screwfix so I have 15W of LED on each track. I can’t find a transformer that will work with them. All 4 LED transformers from 3 suppliers (claimed 16 to 24W output) that I have tried cause flashing of the LEDs indicating current overload. In all cases if I turn it on with less bulbs and then insert the others later it will not flash. So it’s not the steady load that is causing an overload. Or putting the switch in the transformer secondary rather than the primary might lead to a successful start-up.
I think the problem is that these bulbs must have a constant current driver chip such as https://www.st.com/resource/en/datasheet/stcs2a.pdf These have small capacitors in circuit which draw a spike of current when turned on. This inrush current spike may be greater than the transformer’s overload circuit allows, even though it is only 10millisconds long as measured with a scope. See attached traces. These were taken using a LED transformer across 0.1ohm in series with the bulbs so 100mV = 1amp. The second trace is with switching on the secondary side.
Or more seriously, as the transformer voltage rises upon switch-on the LED driver may try to consume more than it’s 12V current. This is because a 15W LED load may consume 1.3A at 12V but, as the transformer output rises through say 6V, it tries to consume 2.6A to deliver 15W, if the bulb’s IC has started up faster than the transformer has. The designer needs more care in designing the transformer so the rise times and time constants are correct. Any overload detector needs a delay circuit of at least 10ms to stop it overreacting to brief high currents. Some LED bulbs may have a soft start feature in them, which may eliminate the inrush, but my oscilloscope indicates that these are not soft start.
One transformer manufacturer stated that the problem was Power Factor in that these bulbs do not have a unity PF. Although a LED transformer has a DC output it is far from smoothed with harmonics from 50Hz to >100kHz. So he said I should aim to keep draw to say 50% of the rating. Interestingly though the one 24W rated transformer I tried was even worse than their 18W one. I noted though that the internal designs were quite different so it’s not just a slightly uprated variant of the same design. I don’t have room for too high a rating transformer.
Another issue is rating. Some transformers rated at say 18W may in reality consume 18W but only deliver 80 - 85% of that, which means that a 15W load is on the limit of what it should deliver. However one manufacturer’s claimed in literature and label that it’s a 18W output transformer but in an email implies 80% of that.
I am tempted to go back to the original tungsten bulb transformers or modern equivalent, if it has failed, as the bulbs do work well with these as they accept an AC input. But my Picoscope software records the RMS voltage as around 6V on such a low wattage load even though the bulbs' brightness ems to be OK. The load is below the minimum these transformer are rated at. A compromise is to fit one tungsten and 2 LEDs to keep it closer to the minimum rating and then 12V is output.
The waveform is horrendous with variable length bursts of many kHz every 1ms. It amazes me that a https://www.st.com/resource/en/datasheet/stcs2a.pdf or similar can cope with such a variable input. Whether it will be reliable long term remains to be seen.
My conclusion is that the manufacturers have designed their transformers to work with the kind of LEDs with a dropper resistor and no IC as these will accept any sort of pulsed input and not have inrush issues. But it’s impossible to tell what sort of LED one has bought by looking at it and the mainstream sellers like Screwfix don’t sell them.
If space allowed I’d just put a 230 to 12VAC standard non electronic transformer to power several tracks at once and ditch all these dodgy inverters.
Any ideas please?
Attachments:
-
forums-5w.png
(26.88 KB)
You must be logged in to download this attachment.
-
forums-15w-with-swiching-after-transformer.png
(15.06 KB)
You must be logged in to download this attachment.
