| MikeHunt79 |
I'm not sure if this is the right section for this, so can a mod please move if it's wrong...
I want to build this... I'm planning on 112.5Hz crossover frequency.
http://sound.westhost.com/project09.htm
This is going to be in a car, so hifi quality isn't really 100% needed.
I have a bunch of 4558D chips on an old car amp I can use, but these look like they are poor quality from what I've read here.
So can someone recommend me an op amp? I know next to nothing about them... |
|
|
| Nuuk |
| If in doubt stick with something like the trusty old 5532. ;) |
|
|
| pinkmouse |
| And the OPA134 series isn't bad either. |
|
|
| Leolabs |
| Something with FETs inputs and able to operate at unity gain. |
|
|
| AndrewT |
Hi,
the 5532 can be used for the input buffers.
tl072 or other FET input opamp works for the filter stages.
The output buffers could be another 5532.
If you use tl072, ensure the parasitic capacitance on the outputs of the opamps is minimised. They don't tolerate capacitive loading. |
|
|
| jethdub |
In reference to previous post, I'm aware of the need to protect chips such as the TL072 from capacititive loading, but just how close to the output pin does the low value R need to be?
Often find myself laying out a circuit and getting nervous about moving the resistor millimetres away.
Also, if a blocking cap is to be used, is there a problem to place it after the load resistor rather than before, as this often allows closer placement of the resistor to the output? |
|
|
| Tincho |
Mike,
I did that project using TL074 .
I'm using it in a work in progress 2.1 PC sound speaker.
I use ground plane layer and 78xx based PSU.
I drive TDA2003 and TDA2030 directly. I'm planning to change these by lm1875.
Sound is good. No noise No hum
Best,
Tincho |
|
|
| fernando_g |
The 4558 is a very dated opamp. Used extensively many moons ago, but long superseded by better devices which will not cost you an arm and a leg.
The posters have already made several suitable suggestions above. I've used every single one of those suggested devices and can tell you it will be plenty, for the application you are looking for.
However, I would add to add to the list the BiCMOS cousins of the deservedly famous TL07x BiFET family....For instance the TLC072. They offer several improvements over the standard family, yet they are quite affordable. check yourself this link.
http://focus.ti.com/lit/ds/symlink/tlc072.pdf |
|
|
| Keruskerfuerst |
| In this case, you should use an DIFET/FASTFET or bipolar input opamp. |
|
|
| MikeHunt79 |
Thanks for the recommendations guys - I only know a little opamp theory, I must admit I haven't a clue what the difference is between a 5532 and a TL074, but I looked up the datasheets which helped a little...
Anyway, am I right in assuming that all these opamps run from +-15v rails? and they use very little power? So would I be able to run 10 op amps with unsinked 7815's? My amps rails are +-24v so the 7815 will only be dropping 9v or so. :)Awesome - I think I'll just copy what you did if that's ok, as I'm a bit overwhelmed with all these different opamps right now. :o Obviously with different value caps and resistors as I want a different crossover point.
Just 1 quick question, if I wanted to adjust the volume of the LP out and HP out independently, could I just wire in a 10k log pot in the same way as a normal volume pot at the LP out an HP out? My amps gains are in the boot(trunk), so they won't be very easy to adjust while driving. ;) |
|
|
| Dr.EM |
| As far as I knew the op-amps should be FET input to ensure high input impedence? This stops the thier input impedence from affecting the predicted response. The 5532 isn't, but can be used for the buffers. I have used OPA2134 chips and been quite satsfied; FET input and unity gain stable. A very low distortion chip too :) |
|
|
| gainphile |
| I built that circuit using TL072 and I cannot fault the sound. Very neutral and costs $2 only :D |
|
|
| juma |
| quote: | Originally posted by MikeHunt79
So would I be able to run 10 op amps with unsinked 7815's? My amps rails are +-24v so the 7815 will only be dropping 9v or so. :)
Awesome - I think I'll just copy what you did if that's ok, as I'm a bit overwhelmed with all these different opamps right now. :o Obviously with different value caps and resistors as I want a different crossover point.
Just 1 quick question, if I wanted to adjust the volume of the LP out and HP out independently, could I just wire in a 10k log pot in the same way as a normal volume pot at the LP out an HP out? |
1. If it's not a big trouble for you, you'll get better results with mildly heatsinked LM317/LM337
2. Of course it's OK, that's what this forum is all about - to share ideas and help each other.
3.Yes, you can use 10k log pots to independently control HP and LP volume, but you'll need to use bigger coupling cap (4.7 uF) on LP output. |
|
|
| OzMikeH |
You;ll need to do something interesting for the power if you run it off 12 volts DC in a car.
Perhaps using 0V 5V and 10V for a pseudo +/- 5V.
you'll need to capacitor couple the input and output audio grounds if you do that. |
|
|
| jethdub |
Juma, I notice on your schematic that you have seperate PS and signal grounds marked. Are these grounds just seprated on the board but fed from a single cable to the star or central grounding point, or are they seperated all the way back to star ground?
Is it normal practice for you to seprate your grounds or only for certain circuits?
This is something I've been reading on recently and it doesn't seem to be at all straightforward so I'd be interested in hearing about your approach. Thanks. |
|
|
| MikeHunt79 |
| quote: | Originally posted by juma
1. If it's not a big trouble for you, you'll get better results with mildly heatsinked LM317/LM337
2. Of course it's OK, that's what this forum is all about - to share ideas and help each other.
3.Yes, you can use 10k log pots to independently control HP and LP volume, but you'll need to use bigger coupling cap (4.7 uF) on LP output. |
I've already got some 7815's handy, so I'll use them for now...
Also, I've discovered my headunit is has 4 channel pre-out, so this has changed things a little... I was thinking of using channels 1 & 2 for L&R above 112Hz, and use channels 3&4 summed together below 112Hz...
This should make things simpler, as I'll need less filters and maybe less input and output buffers (I think)...
I'm still getting to grips with eagle, I only got it today, and I'm hoping to put together a schematic, which should show more clearly what i plan to do with the extra channels....| quote: | Originally posted by OzMikeH
You;ll need to do something interesting for the power if you run it off 12 volts DC in a car.
Perhaps using 0V 5V and 10V for a pseudo +/- 5V.
you'll need to capacitor couple the input and output audio grounds if you do that. |
This won't be a huge problem, as I already have an amp with +-24v rails.
I've adapted this circuit a little: http://sound.westhost.com/project05.htm
This should work shouldn't it?
Also, does anyone know where I can get the eagle library which has the 1n5818 in? I ended up using a renamed 1n5400 for now... |
|
|
| MikeHunt79 |
Ok, I've ordered a load of TL072's. :)
I've got 8 pin sockets handy, so I can use them and use my 4558's while I'm waiting for the TL's to arrive.
Anyway, I'm stuck... I've found a voltage summer here:
http://users.cscs.wmin.ac.uk/~wooda/components/opamps/
Now, the part which concerns me is "R5 should be parallel combination of R1-4 (R1-3 including source impedances)."
I'm not really sure what the source impedance is, but would using input buffer solve this? Or could I just get rid of R5 and have an inverting summer?
I'll post the schematic so you can see what I'm trying to achieve... |
|
|
| MikeHunt79 |
here's the right channel (left channel is the same...):
Also, I'm only going to go for volume on the sub channel, using an output buffer.. I just haven't added this to the schematic yet.... |
|
|
| jethdub |
| I believe that as you are using TL072 amps the resistor on the non-inverting input isn't necessary as Dc offset is not such an issue with FET input amps. |
|
|
| MikeHunt79 |
| quote: | Originally posted by jethdub
I believe that as you are using TL072 amps the resistor on the non-inverting input isn't necessary as Dc offset is not such an issue with FET input amps. |
Exellent, so I can just lose R5 and it will work ok? Thanks. :)
EDIT: Would it be worth adding input buffers before the voltage summer? |
|
|
| MikeHunt79 |
Ok, updated sub schematic...
Does this look ok? |
|
|
| poobah |
As long as both inputs are being driven by like sources, as in output impedance, there's probably little need to buffer them.
:) |
|
|
| AndrewT |
| quote: | Originally posted by MikeHunt79
http://users.cscs.wmin.ac.uk/~wooda/components/opamps/
Now, the part which concerns me is "R5 should be parallel combination of R1-4 (R1-3 including source impedances)."
I'm not really sure what the source impedance is, but would using input buffer solve this? Or could I just get rid of R5 and have an inverting summer? |
Hi,
the non-inverting input needs a resistor to ground on the input to supply the matching input offset current that the FET opamp expects to see.
This offset current is very small and the opamp will tolerate a significant error in the matching and still give an output with acceptable output offset.
But, the solution is easy.
Add A DC blocking capacitor to the input and take the resistor to ground. This way the filters and opamps are protected from DC coming from the source and the input offset current is supplied by the signal ground connection.
I recommend the inverting opamp as a summer. But remember the gain is defined by the value of all the parallel input resistors and their source resistances. |
|
|
| juma |
| quote: | Originally posted by jethdub
Juma, I notice on your schematic that you have seperate PS and signal grounds marked. Are these grounds just seprated on the board but fed from a single cable to the star or central grounding point, or are they seperated all the way back to star ground?
Is it normal practice for you to seprate your grounds or only for certain circuits?
This is something I've been reading on recently and it doesn't seem to be at all straightforward so I'd be interested in hearing about your approach. Thanks. |
Hi Jet,
if you look at the bottom view pic you can clearly see the ground arrangement (black wire is power gnd). It's a bus ground scheme that leads to amps' star ground point with 10 Ohms loop-breaker. You can see many good examples in Pass Labs forum and passdiy.com (http://www.passdiy.com/pdf/zen-v4.pdf for instance) |
|
|
| juma |
| Mike, your schematic from post #17 won't work: IC2 should be 7915 (negative voltage regulator) |
|
|
| MikeHunt79 |
| quote: | Originally posted by poobah
As long as both inputs are being driven by like sources, as in output impedance, there's probably little need to buffer them.
:) |
Yep, they are all from the same source... 4 channel headunit. :) I'll skip the input buffers then. | quote: | Originally posted by AndrewT
Hi,
the non-inverting input needs a resistor to ground on the input to supply the matching input offset current that the FET opamp expects to see.
This offset current is very small and the opamp will tolerate a significant error in the matching and still give an output with acceptable output offset.
But, the solution is easy.
Add A DC blocking capacitor to the input and take the resistor to ground. This way the filters and opamps are protected from DC coming from the source and the input offset current is supplied by the signal ground connection.
I recommend the inverting opamp as a summer. But remember the gain is defined by the value of all the parallel input resistors and their source resistances. |
Yep, I've used 10k resistors so the sum of both voltages should go to the output, without the output voltage being multiplied or divided...
Ok, so If I have a 100nF cap between the signal and ground on both channels 3 & 4 before they reach the summer resistors, that should get rid of the DC offset problem?
I'm not quite sure what you mean tho when you say "take the resistor to ground"... do you mean I should have R7/R8 inline with the 100nF cap between sugnal and ground (instead of it being in-line with the signal)?
| quote: | Originally posted by juma
Mike, your schematic from post #17 won't work: IC2 should be 7915 (negative voltage regulator) |
Ooops, I've changed it to a 7915 now, thanks. :)
Finally, for the output buffer, I looked high and low and couldn't find a 20k pot. Could I use a 10k pot if I changed all the fixed resistors in the output buffer (r9,r10,r11) to 5k? |
|
|
| AndrewT |
Mike,
go back and look at the opamps link you posted.
initially I'll use the last three diagrams for explanation.
Take the last one. High pass non-inverting unity gain S&K 2pole filter.
The first capacitor acts as the DC block and ensures that any DC present in the input signal does not pass to the output of the filter.
The two diagrams above show the input resistively coupled to the opamp input pin.
These two, the summer and the low pass, will send DC in the input signal to the output.
They both require a high pass filter (=DC block) on the signal line.
A series capacitor that is big enough to pass ALL audio signal can be inserted between source and the diagrams. Somewhere around 1uF to 10uF should be about right for most filter circuits and buffers. A low impedance buffer may require the series C to be as high as 100uF (Rane, as an example, suggest 220uF for their balanced lines). The resulting diagram will look like the AC coupled amplifier but without the biasing and C2 on the non-inverting pin.
Now the summer. If r1=r2=r4 (for a two channel summer) then the gain for each channel is 0db (=1times). If the signal on the two inputs is identical in voltage AND phase then the output is twice the single input voltage i.e. output voltage =V1+V2. This is due to the two input summing resistors being in parallel and presenting a net source resistance of 5k (=10k//10k) the gain becomes 10k/5k (R4/Rsource) = +6db (2times).
Now the resistor to ground ambiguity.
The high pass filter is a capacitor in series with the signal and the Resistor takes the output of the capacitor to ground. The feed to the next stage is taken from the junction of the C&R.
Look at non inverting amplifier (diagram4). R3 is the resistance seen by the non-inverting input pin and the formula says that this resistance should be the parallel sum of the two resistors attached to the inverting pin. For a gain of 2, both these resistors, R1=R2=10K, have a parallel sum of 5k. The formula says that R3=R1//R2=5k.
If you fit a DC blocking cap in front of this resistor then the effective resistance seen by the non-inverting pin is infinity due to the perfect capacitor. Instead what we can do is take the resistor from input pin to ground and take a second feed from the non-inverting pin to the DC blocking cap and then to the signal input terminal. The opamp still sees 5k resistance AND the signal has a DC blocking cap in it's route. |
|
|
| jethdub |
| Just wanted to check something that I realised I ws unsure of looking at Mikes schemos. The first filter stage is fed from the summing amp, is the source impedance then the 10k summer feedback resistor in parallel with the opamp output? |
|
|
| AndrewT |
| quote: | Originally posted by jethdub
The first filter stage is fed from the summing amp, is the source impedance then the 10k summer feedback resistor in parallel with the opamp output? |
Yes, the pair of impedances add in parallel.
The 10k feedback resistor is usually ignored due to the low value of opamp output impedance. This Zout is probably <10r and may be a lot lower. |
|
|
| jethdub |
| Thanks, I knew that really, just that whilst on the learning curve and with so much information to absorb at times things become a little confused... |
|
|
| MikeHunt79 |
This has been on the back burner for a while, but I haven't forgot about it...
I'm about to order the parts now, and I've already got loads of metal film resistors, but I haven't a clue what sort of caps to get...
All I know is that electrolytic ones can't be used, as they are not bipolar... so what caps would be best out of these?
Aluminium Electrolytic (7,290)
Carbon Aerogel Foam (67)
Ceramic Disc & Plate (1,432)
Ceramic Multi-layer (9,275)
PET (96)
Paper (60)
Polycarbonate (14)
Polyester (1,821)
Polyphenylene Sulphide (331)
Polypropylene (1,533)
Polystyrene (144)
Related Products (4,048)
Silvered Mica (216)
Tantalum Electrolytic (4,298) |
|
|
| Eva |
Low cost op-amp alternatives:
- NE5532 is great for driving long signal lines or low impedances. It has a low input impedance so it's not recommended for filters.
- LM833/LM837 are good overall, except in high impedance filter circuits.
- TL07x is good for high impedance filters due to the FET inputs resulting huge input impedance. |
|
|
| BWRX |
| quote: | Originally posted by MikeHunt79
so what caps would be best out of these? |
Ceramic Multi-layer - C0G or NP0 types only
Polyester
Polypropylene
Polystyrene |
|
|
| AndrewT |
| quote: | Originally posted by BWRX
Ceramic Multi-layer - C0G or NP0 types only
Polyester
Polypropylene
Polystyrene |
and if you can still obtain them add polycarbonate to the list. |
|
|
| MikeHunt79 |
Thanks guys, I've gone for some ceramic ones.
Eva, I've gone for TL072's, by the looks of it they should work. :) |
|
|
|
