I have made a three way active system.
The bass uses Linkwitz filtering fed into a bridged pair of TDA1514.
The bridging is done by buffering and then inverting using NE5532. Then feeding inv and non-inv into the two power op amps.
It has worked nicely for a few months.
For the second time I am having to replace one of the power-amps which has stopped working.
The previous time it stopped, it did so pretty spectacularly (smoke etc.) and the trigger was to inadvertently remove the lead from the crossover and thus have a floating lead feeding into the Linkwitz filter + bridge (all in the same box, behind the speaker).
The power supply sits under the speakers, so there is about 12 inches of cable snaking up to this amp.
Any general suggestions to improve stability in a setup like this?
The bass uses Linkwitz filtering fed into a bridged pair of TDA1514.
The bridging is done by buffering and then inverting using NE5532. Then feeding inv and non-inv into the two power op amps.
It has worked nicely for a few months.
For the second time I am having to replace one of the power-amps which has stopped working.
The previous time it stopped, it did so pretty spectacularly (smoke etc.) and the trigger was to inadvertently remove the lead from the crossover and thus have a floating lead feeding into the Linkwitz filter + bridge (all in the same box, behind the speaker).
The power supply sits under the speakers, so there is about 12 inches of cable snaking up to this amp.
Any general suggestions to improve stability in a setup like this?
Well, I dont know how much power those things put out, but do a search on "FRD Consortium" and look at the Linkwitz Transform simulator spreadsheet. It will let you see how much power is required for your application. I am guessing it will be VERY large. I use a 380W amp for mine, and it is barely enough.
You might also want to invest in some measurement software and a mic to measure your in-room response. I measured mine, and even with my roll-off starting at 70Hz due to the small sealed box, I was able to get near flat bass response. I really only use the LT to lower the Q of the box.
-Paul
You might also want to invest in some measurement software and a mic to measure your in-room response. I measured mine, and even with my roll-off starting at 70Hz due to the small sealed box, I was able to get near flat bass response. I really only use the LT to lower the Q of the box.
-Paul
oops,
What I was getting at was that you are drawing too much current from your chips.
Maybe try making another amplifier, and doing one amp per voice coil or one amp per driver etc.
Or try a bigger amp
-Paul
What I was getting at was that you are drawing too much current from your chips.
Maybe try making another amplifier, and doing one amp per voice coil or one amp per driver etc.
Or try a bigger amp
-Paul
I am not sure at the moment if the NE5532 is unity gain stable (it's brother the 5534 isn't). So are you sure that you don't have any oscillations on one branch of your bridge ?
Just an idea in case that it is always the one fed by the inverting stage that blows.
Regards
Charles
Just an idea in case that it is always the one fed by the inverting stage that blows.
Regards
Charles
phase_accurate said:
I thought it is, but wanted to check the datasheet first. They
don't actually say anything about whether it is unity gain stable
or about any minimum closed-loop gain,
and there are no phase curves in the frequency plots.
They only thing they say is that it is internally frequency
compensated, which, considering the absence of further info,
I assume should be interpreted as that
it is unity gain stable, although it does not necessarily mean
that.
Hi Charles,
The NE5532 is the internally compensated & unity gain stable dual version of the single NE5534 which is stable at gains >3.
What else can make a bridge-clone unstable?
Let's assume that the problem is unlikely to be in the NE5532 inversion.
What about the following:
1. I have used a zobel (100nF 3R) for the output of each half of the bridge. Is that correct?
2. When using parallel bridging the circuits in the forum put .22R (or thereabouts) in series with the output of each power amp. Would that help add stability in this situation?
3. As I said the topography of my active speakers (i.e. very short speaker leads, but power supply in the base of the speakers) means 12-15 inches of power cable. I have put smoothing capacitors in the base unit, the gain clones each actually have 47000uF for each rail. Would a resistor in the 20V lines prevent the power supply leads becoming a source of instability? If so what sort of value?
4. I have put blocking capacitors + resistor to earth at the between the output of the NE5532 and the input of the TDA1514 (as per datasheet). I am assuming that this well help to protect the bridge against offset voltages being amplified and presented to the load (speaker).
5. Both gain-clones are running off the same "umbilical" lead. I have twisted together the +ve and -ve for each TDA respectively within the casing. Anything else to be careful about here?
It would be very helpful to get this sorted. Thanks very much for all efforts so far
Let's assume that the problem is unlikely to be in the NE5532 inversion.
What about the following:
1. I have used a zobel (100nF 3R) for the output of each half of the bridge. Is that correct?
2. When using parallel bridging the circuits in the forum put .22R (or thereabouts) in series with the output of each power amp. Would that help add stability in this situation?
3. As I said the topography of my active speakers (i.e. very short speaker leads, but power supply in the base of the speakers) means 12-15 inches of power cable. I have put smoothing capacitors in the base unit, the gain clones each actually have 47000uF for each rail. Would a resistor in the 20V lines prevent the power supply leads becoming a source of instability? If so what sort of value?
4. I have put blocking capacitors + resistor to earth at the between the output of the NE5532 and the input of the TDA1514 (as per datasheet). I am assuming that this well help to protect the bridge against offset voltages being amplified and presented to the load (speaker).
5. Both gain-clones are running off the same "umbilical" lead. I have twisted together the +ve and -ve for each TDA respectively within the casing. Anything else to be careful about here?
It would be very helpful to get this sorted. Thanks very much for all efforts so far
- Status
- Not open for further replies.