UcD400 Driving A 1/2 Ohm Ribbon?

[valveitude]

After nearly going blind reading the monster 180/400 Q&A threads over several sittings I believe this will work, but wanted to run this past you guys before I fire up the soldering Iron.

My ribbon project is nearing completion and I need to start thinking more on how I'm going to drive it. One of the options I am seriously considering is to drive it directly with a UcD400. It has been stated several times that it's stable into a 1 ohm load, but is it into a 1/2 ohm load? My understanding is that it has a very robust current limiting, and if it is stable into such a low load, it would provide around 50 watts into 1/2 ohm. Since it would only need to swing around 5 volts at "full power", I plan to bypass the buffer stage and tap into the modulator directly with a 1:1 isolation transformer which would be driven directly from my active crossover. The gain of around 4.5 of the power section works out to be near ideal.

The only niggle I can identify with my plan is the output impedance rising from 20 mOhm at 1 khz to 150 mOhm at 20khz. Low enough to not be a consideration with a normal load, but with a 500 mOhm load, I expect a slight droop in frequency response at the top end. I'll have to "suck it and see", but if it is noticeable I can "tilt" it up in the crossover circuitry..unless anybody has a better plan.

Well, thats the plan...thoughts?
Casey

[ericpeters]

I have been using Zappulse 2.2 amplifiers for my 3 to 4 ohm Apogee Duetta Signatures, these work very well. However when we attached the same amplifier to a .5 Ohm Scintilla the sound got very dull. The Zappulse together with the Speaker works as a Low Pass filter. With a 4 ohm speaker the cutt off is way above the audibel range (probably around 50.000hz or higher) but with a lower impedance the cut-off got in the Audible range.
I do not know whether this is a Class-D typical or a Zap pulse typical effect.

[valveitude]

Hi Eric,

I do not know whether this is a Class-D typical or a Zap pulse typical effect.

If I had to guess, I would say its the effect of the lowpass filter on the amp. I have no experience with ZapPulse, and only a little with Hypex, but based on the output impedance curve of the UcD400 in the datasheet I cobbled together a sim on spice. The results were that it was down just shy of 1dB at 20khz in the not-so-accurate sim. If the Zap Pulse carrier freq is much lower than Hypex, I would guess that this "droop" would be more pronounced due to the required higher inductance of the output coil...just guessing.

Casey

[Tim__x]

Keep in mind, that rise corresponds to about 1uh of output inductance, the Hypex literature it telling the truth when it says most conventional amplifiers don't achieve that.

[ackcheng]

I tried UCD400 with my Raven 2.0 which goes down to about 2.5 ohms. In my set up, I drive the tweeter directly (active) but as soon as I switch the amp on, the over current protection kicks in. I need to place a 2 ohm resistor in series with the tweeter for it to work.

[valveitude]

Well I did what I should have done in the first place...went to the source and emailed Hypex with my questions.

The money quotes in my exchange with the man himself, with permission, Bruno Putzeys...

There should be no issues with stability or frequency response driving a low-ohmic load, but distortion will definitely go up (by around a factor of
8).It's difficult to predict to what extent this would be a sonic problem
or not.

To put it more exactly, the distortion is caused by current,
but shows up as voltage. If distortion at 10W/4ohm is 0.01%, that's 630uVrms
unwanted signal caused by 1.6Arms. If you were to drive 1.6Arms into
0.5ohms, that's 1.28W or 0.8Vrms. In absolute terms distortion is the same
630uVrms but against 0.8V that's 0.08% THD. Power wise you've got 8 times
the distortion for one-eighth the power. So, you can model the amp as an
ideal amp with a nonlinear impedance in series. The lower the load
impedance, the more significant the series impedance becomes.

So thats that then. Though it would work with no stability issues, the distortion penalties are just to high.

Plan B)..I will wind an 8:1 toroidal autoformer. This will present a 32 ohm load to the amp with a half ohm load. I only need 30 watts or so for ear bleeding levels (the projected efficiency of the ribbon is 98 dB/W), and this will give me 50W. As a major bonus, I will potentially decrease the distortion by a factor of 8 ...distortion tracks current. Admittedly some, if not all, of these gains will be offset by the distortion added by the transformer, but hey, it's worth doing.

Casey

[hanswolek]

Hi,
i believe based on datasheets and forums that on UCD400 current limiter activates on peak current of 20-21A. RMS current therefore is about 14A. That should result in about 100W into 0.5 ohm, and 200W into 1ohm and 400W into 2 ohm (and about 530W into 3 ohm, and 400 into 4 ohm as well as no current limiting occurs then).

Would appreciate if someone can confirm this.
Thank you.

[Yves Smolders]

Let me tell you a story about that one... When building my amp I hadn't noticed that my speakerposts were shorted against my case. Amp comes up, doesn't go in protect. Speaker is playing but softly, it goes into clipping pretty rapidly, and some electronics on my module is "singing" along with the music. I was wondering for minutes what was wrong, thought I made a mistake in the input signaling. Effectively, I was pushing a UCD400 into 0.x ohm load, whatever the case was doing, with an 8 ohm speaker parallel on that. No smoke, no smell. Repaired the situation and modules are fine. Tell me what other amplifier you can drive into a short circuit and not going up in flames?

These modules are not rock solid, they are granite solid.

[ViennaTom]

@valvetude:

You could also try modding the UCD400 to better fit to your impedance (instead of impedance matching XFMR adding new distortions)
Just an idea: UCD400 is recommended for >=1Ohm load impedance. It operates in voltage mode, e. g. control loop feds back output voltage. The filter stage basically consists of L and C. So, from the control loop's point of view, L -> (C par. 1Ohm) looks the same as L/2 -> (2*C par. 0.5Ohm). This way you could boost slew rate into your low load impedance. As a negative side effect, the idle power losses would most likely increase due to higher triangle current through the half bridge FETs..