Lets talk Linkwitz Transforms (1 Viewer)

[Seth_L]

I downloaded the linktranadv29.13.xls spreadsheet and have been playing with it.

I have 2 issues.

1) why +12V and -12V? This requires a fairly elaborate powersupply. I would think using different op amps would solve this problem, but that's not really discussed either.

2) Is the input buffer needed or not? I can't find any information on that part.

Seth

 

[Seth_L]

Upon further thinking I can't see why such wide voltage range op amps are suggested. Line level signals rarely break 2 Volts RMS. There's no reason to accomodate a 24V swing. It seems to me that something like a MAX4167 would be exactly what's needed. It'll work with just GND and 5V. That'll give it an output swing of ~4V which I would think is more than you need in an LT. It has a high slew rate, good rejection from the power supply, and low noise.

Am I missing something?

Seth

 

[Isaac C]

Generally, input/output signals should be 3V (or more) clear off the op-amp's supply rails. So if you're sure your input/output amplitudes are well below 12V, you could use a lower supply voltage(s) (keeping in mind the minimum 3V allowance and the particular op-amp's needs).

But if you were adding circuits to an existing equipment, such as an op-amp-based preamp, why not save a few $ by using the existing supply rails

Isaac

 

[Seth_L]

Isaac,

Where did you come up with that 3V rule? I've never heard of it before.

I'm not adding circuits to existing equipment, hence why I don't want to buy a transformer with a center tap, and a positive and negative supply. I want to use GND and only a positive supply.

Seth

 

[Isaac C]

3V of allowance is just a rule of thumb. You'd see in op-amp datasheets what the maximum output voltage swing is before it clips. For most op-amps, the maximum is 2V to 3V away from the supply rails (unless otherwise specified).

I don't want to buy a transformer with a center tap

No problem, you could use a transformer with only one secondary and you could still get +/- supply rails. Here's an example circuit.


Isaac

 

[Seth_L]

Isaac,

There are a lot of op amps that can get within .5V of their rails, or even closer. If the amp I'm driving has a 1.5Vrms (or less) input sensativity and an input impedance of 6k ohms. I should be able to get away with much lower voltages on different op amps. That I'm confident of. What I'm unsure of is if I can use an op amp that only needs 0 and 5V (giving me more than 4Vrms on the output), or if I need to give it -2.5 and +2.5V (also giving me 4Vrms on the output).

Seth

 

[Jonathan M]

Seth,

Assuming you are right about the LT not needing more than a +-2V swing (This may not be enough IMO - remember to factor in the gain in the LT), you can get away with a single supply opamp.

You need to make sure that you introduce an artificial earth line at half the supply voltage to reference the signal to.

This can be fairly easily done (Just use a voltage divider - resistors to both the +ve and -ve lines). I believe that there may be such a circuit on Rod Elliotts site for use in a vehicle. Try here for his circuit design. You won't need as much filtering, ofcourse, as you won't be using it in the car will you?

I suggest the whole project will be best using a 12V supply at minimum. This will allow TL072's or similar to be used.

As was said above, however, you don't need a centre tapped transformer to get +/-12V. Check out Project05 at ESP for how to do this.

Hope this has given you some ideas.

 

[Janne Ahonen]

Seth,

If you use single-supply, then you'll have to modify LT circuit so that everything is biased at VCC/2. That means that input and output signal must be AC coupled.

LT circuit must be fed from low impedance source, otherwise it won't work as designed. So input buffer is quite mandatory, in most cases.

Regards,
Janne

 

[Seth_L]

Jonathan,

Wouldn't this be easier?

http://sound.westhost.com/project43.htm

Unless I've bungled the computation 12dB of gain means about 4x the voltage level. Looks like I need about a 6Vrms output swing.

I'll take a look at the TL072 or something similar.

Seth

 

[Jonathan M]

Yep - that's the idea. The opamp is just there in case there is a variance in current between the two supply rails. You shouldn't have too much problem with the LT circuit, as the currents will be small anyway. May pay to measure the two rail voltages after you've built it to see how they stack up - you can always alter the resistances a little if need be.

Make sure you AC couple the input and output of the circuit (If you are using Rod's schematic it'll be fine).

What are you using for the supply - is it regulated or just cap filtered? I suspect it is a DC plug pack? Remember that any 60Hz noise introduced will be amplified a fair amount.

 

[Seth_L]

Jonathan,

I haven't decided on anything yet. I'm just trying to get an idea on what to use to built the LT I'm thinking of for the sub I have in mind. I want to build a sealed Monster.

I would be putting a low loss voltage regulator at the front of the LT's box's circuitry most likely. Then I would probably power the LT box with a wall wart.

Janne,

Thanks for the information. I don't feel like re-biasing the LT to be honest. I originally didn't think that I could use a shifted ground to get the positive and negative voltages. Hence my questions.

Seth

 

[Dan Wesnor]

Seth,

If you have a question about R. Elliot's circuits, why not ask him? I understand he's a pretty responsive guy.

 

[Seth_L]

Dan,

The question isn't really about Rod's circuits. It's more a fundamental question about the design of a Linkwitz Transform. I suppose I should break out my old college textbooks and actually solve it.

Seth

 

[Dave Milne]

Seth,
You're right that you can get op amps capable of swinging to within about 0.5 V of the rails, but I wouldn't try running a single 5V supply. That's just not much headroom. I would use Rod's dual-regulator supply or a variation with CT transformer and regulators (CT's are a few bucks from DigiKey or Newark). That way you can avoid the input and output caps. We're talking subs here, so low frequency response is paramount. You'll spend some serious cash on two large high-quality mylar, polypropylene, or teflon caps if you go single-rail. Remember you can only count on 5-10K input impedance on the power amp, so it's going to take upwards of 10uF to keep the pole down in the 5 hz region. Power amps typically have DC-servo loops so most are immune to a little DC offset at the input. If you're really worried about DC offset, you can "sum in" a correction at the last stage using a pot and resistor.

With +/-12 or +/-15 V, you are free to select the best-sounding amps like the AD797 (watch layout, bypassing, and feedback caps for HF stability) or some of the TI/Burr-Brown devices. The TL072 is mid-grade at best.

 

[Seth_L]

Dave,

I understand the issues with going single rail, but what about going +/-6V? It no longer has any single rail issues, but doesn't need as wide of a swing on the power supply.

I'm not sure I follow you on the 10uF cap myself. Following the schematic in LT spreadsheet you end up with a low pass filter on the output, not a high pass filter. Or were you still talking about single rail applications?

Also, does anyone really know how wide of a voltage swing is really necessary for a LT to properly function? I'm sure I can find an appropriate op-amp, but I need to know how much voltage gain is needed for the LT to properly function.

Seth

 

[Dave Milne]

Seth,
+/- 6V might work, but I think the op amps perform best at full rated voltage. There's not much reason to run them at lower voltage, assuming that you're going to have to buy a transformer anyway. I suppose you could save a regulator if you regulate first and then split with an artificial reference, but 7815/7915 regulators were about 40 cents each, last time I checked.

The cap discussion was related to single rail "artificial reference" schemes. You need the AC coupling caps because the internal reference is not ground. The output blocking cap then forms a high-pass filter with the input impedance of the power amp. Same on the input, but since you have control of the LT circuit input impedance, it's less of a problem.

does anyone really know how wide of a voltage swing is really necessary for a LT to properly function?

This depends on your power amp sensitivity. Usually it's a few volts for rated output. This should be in the amp specs somewhere. For example, the Bryston 4BSST is switch-selectable for 2.3V or 4.6V. Remember that 4.6 Vrms is 13 V peak-to-peak.