TOP-switch HX Family Drops Too Low in Frequency
I am using the TOP261YN (66kHz) and the TOP258YN (132kHz) in two different flyback supplies for audio amplifiers. I have used these for some years with overall good results, using primary side regulation from the bias winding according to the application notes.. However, the response at light load does not meet spec. The spec sheet says the duty cycle and frequency should fall off at light load to a minimum of 30kHz. Instead, the duty cycle remains fairly close to full length (3uS on pulse) and the frequency on both devices drops to 9-10kHz. This causes a very annoying mechanical whine in the flyback transformer, which has vacuum impregnated varnish to the usual standards.
The older GX series behaves better in the same application. Duty cycle reduces to 0.7uS before frequency begins to drop. Thus, even if the frequency does drop into the audible range the impulses to the transformer will be much smaller and less noisy.
Please advise whether this can be corrected. Otherwise the spec sheet should be amended.
Comments
Hello,
The TOPSWitch HX has four modes of operation:
1. MCM - Muilti cycle modulation - Burst mode operation
2. Low Frequency Mode - Approximately 30kHz operation
3. Variable Frequency mode
4. Full Frequency Mode (132kZ/ 66kHz operation)
This multi mode operation enables design of power supplies that maintain very high efficiency down to very light load.
From your description, it appears that the power supply is running in MCM mode due to light load. typically a varnished transformer should not produce audible noise however if you believe you are finding that under certain conditions, the unit produces audible noise at light load, redesign of transformer so as to reduce the flux density helps. Kindly review.
It is also useful to replace any ceramic capacitors in the snubber clamp with film type capacitors as often the source of noise under burst mode operation is audible noise from ceramic capacitors due to electrostriction.
Hope this helps. Please review page.9 and figure.9 of the datasheet where this is explained.
http://ac-dc.power.com/sites/default/files/product-docs/tophx_family_da…
Mr Quilter:
Have seen the same. Best cure is to load it more heavily, or use a smaller device so load is a higher percentage and the unit stays out of MCM. The latter is not good if the load is sometimes high but loaded/unloaded ratio is large. Loaded/unloaded ratio is often a problem with SMPS as you know.
Noise is bad for audio, but usually the fan in QSC amp (if supplied) should drown out most mechanical noise.
J Tiers (former Ampeg/Audio Centron)
Hi,
the audible noise can sometimes result from the resulting average frequency during MCM operation matching the frequency of mechanical resonance of the transformer structure[ Core + Bobbin]. This tends to happen more easily for long cores. change of core/bobbin form factor often helps to resolve this.
In some situations, it helps to put a suitable glue between the core and the bobbin after winding the transformer and also in the air gap while inserting the cores on the bobbin. This helps to damp the vibration and reduce the audible noise to levels where it is imperceptible.
Sometimes the problem is stable low frequency operation (10kHz), sometimes bursts of several cycles with gaps, at approx. 5K repetition rate, especially at 230Vac vs 120Vac. I modified the feedback poles and zeros +/- 2X without significant effect. The problem still seems to be that the on-pulse never becomes short enough to allow light load regulation without the frequency dropping too low and/or skipping pulses. The minimum length of the on-pulse appears to be determined inside the chip so I don't see how to affect that variable externally. The older GX series minimum on-pulse can go approx. 4X shorter, so it doesn't need to drop so low in frequency at light load.