Thanks for the replies. I do like the output transformer idea. I think I might prototype several variations and see what kind of results I'm getting from each method, regarding noise and sound quality, before building a bunch of them. I do know already that the completely transformerless SCA T15 circuit sounds wonderful to my ears, but, I have heard some interesting "input transformered" Five Fish SC1 sound recordings as well. The output transformer only circuit is an intriguing mystery I'm looking forward to listening to. It seems like most mic preamp designers will leave off the output transformer in favor of keeping the input transformer when designing with the distortion and coloration in mind. I've also heard murmurs that the sonic effect of an output transformer is more "subtle" than that of an input transformer in a preamp circuit.
that 1512 or similar preamp
- Thread starter rolo95
- Start date
Help Support GroupDIY Audio Forum:
Regarding adding an output transformer, I assume it comes right after the 1512's output. David at Cinemag advised me on the need to terminate the secondary of an input transformer with a stepped up turns ratio, like the CMMI-8PCA and CMMI-10PCA to a resistor of the right value, using a formula he described, with the THAT 1512 circuit. Is there a similar need for series resistance when using an output transformer with the 1512? Any theories involved would be good for me to learn, I have some text to refer to if necessary.
I'm looking at the EA 2503, which has a primary impedance of 75 ohms, and a secondary impedance of either 75, 300, or 600 ohms. I'm also looking at the EA 1166-500 which has a primary impedance of 200 ohms and a secondary impedance of either 150 or 600 ohms. I don't know if these are appropriate for the circuit, I just know I like the 2503 based on the API 312 circuit, and the 1166-500 just looks like something I would want to try. I seem to have gathered that 600 ohms is the desired standard line output impedance? Sorry I'm still pretty new to electronics theory but want to learn.
Thanks for any help and getting me in the right direction.
I'm looking at the EA 2503, which has a primary impedance of 75 ohms, and a secondary impedance of either 75, 300, or 600 ohms. I'm also looking at the EA 1166-500 which has a primary impedance of 200 ohms and a secondary impedance of either 150 or 600 ohms. I don't know if these are appropriate for the circuit, I just know I like the 2503 based on the API 312 circuit, and the 1166-500 just looks like something I would want to try. I seem to have gathered that 600 ohms is the desired standard line output impedance? Sorry I'm still pretty new to electronics theory but want to learn.
Thanks for any help and getting me in the right direction.
Meathands
Well-known member
monkeyxx said:
Great idea. We can type and theorize all day but in the end it's your ears the circuit has to answer to. Well, your ears and the laws of physics. This is where prototyping to evaluate noise can get tricky. A great deal of the noise performance of a low-noise circuit like this will be determined by your shielding and PCB layout. But while you're proto-ing, you may want to try an approach with a ground plane a la Jim Williams.
There's a wonderful discussion of breadboarding with a ground plane in Linear Tech's AN47, which as a hole is one of the greatest things about electronics I've ever read.
monkeyxx said:
Could be. Without a pad after the output transformer, you'd have a hard time driving it into color land without also driving your converters into shittsville.
If you want something clean and cheap with an input transformer, but want to avoid the problems JR brought up, how about a simple IC opamp? Jensen has a couple schematics designed to flatter their input transformers:
http://www.jensen-transformers.com/as/as017.pdf
http://www.jensen-transformers.com/as/as018.pdf
You could get as fancy as you want with the opamp choice and balance the output with a transformer or line driver IC.
Meathands
Well-known member
Man, David is great. Without being privy to your conversation, I would guess that he was concerned with making sure that the input impedance of the preamp remained high enough. Transformers ideally have no impedance of their own. Rather, each side "reflects" the impedance of the other by a degree of the turns ratio squared. So, say you're using a 1:8 input transformer. You need the impedance on the secondary side (Zs) to be high enough that is Zs/8^2 is at least 1k (typical mic preamp input Z).monkeyxx said:
When using an output transformer, you simply want to make sure that the output impedance remains low and that you are presenting a high enough input impedance to the 1512. More on this in a sec... There are other concerns with using an output transformer, such as damping ringing with a Zobel network (resistor and cap across the secondaries). I don't know much about this, so I'll let someone else chip in.
monkeyxx said:
The 600 Ohm standard is a telecom standard that was used in some old gear but is obsolete now regarding pro audio. Forget it. In broad terms, you want the output impedance of almost any device to be as low as possible, meaning it can provide as much current as could be needed. When choosing the turns ratio for the output transformers the trade off is voltage for impedance. For every degree of output level you gain (voltage), you lose that amount of current. Or to put the same thing another way, the output impedance increases by the square of the turns ratio.
You also want to make sure you are loading the stage driving the transformer (1512 in our case) properly. The 1512 datasheet recommends a 2k minimum load. To make sure you're providing this, look at the impedance on the secondaries and the turns ratio of the transformer. Then it's the same calculation: square of the turns ratio. If you use a Zobel network, the value of the resistor will dominate the impedance on the secondaries and you can use that for your calculation. If you leave the transformer un-terminated, the input impedance of the next stage (your converters, mixer, etc.) will determine it.
If you're looking to get a little color out of your output xfmr, I would try a 1:1 turns ratio so you can drive it harder while sending less level to the next stage.
Thanks a million! That's some great information. Yes the square of the turns ratio is the same formula Dave mentioned, but your explanation gives me a broader understanding of what's going on. I need to do some reading on impedance. I've just only recently gotten the hang of Ohm's Law, and it felt like the clouds were opening. I think I have almost enough information now to start a build.
I'm looking at the SCA T15 schematic as my reference, since I really liked that unit, and actually what comes before the balancing line driver stage is the OPA134PA buffer. I looked at the datasheet and could not find anything like "minimum load." I think I read that this buffer decreases noise when turning the gain trim pot. What kind of load would the OPA134PA need?
I'm also wondering why the API 312 schematic has no apparent load resistors between the 2520 and either of the transformers? With the exception of the zobel network on the input transformer, of which the resistance is much lower than theorized using the turns ratio squared formula. I've also been told the zobel network is even "optional" on the CMMI-8PCA in particular. I'm wondering how the microphone input impedance and line output impedance are set on the 312. I know the -20dB input pad has an impedance of about 1600 ohms but it's not always engaged. Maybe understanding this will give me a better grasp of the theory at hand.
I'm also wondering why the API 312 schematic has no apparent load resistors between the 2520 and either of the transformers? With the exception of the zobel network on the input transformer, of which the resistance is much lower than theorized using the turns ratio squared formula. I've also been told the zobel network is even "optional" on the CMMI-8PCA in particular. I'm wondering how the microphone input impedance and line output impedance are set on the 312. I know the -20dB input pad has an impedance of about 1600 ohms but it's not always engaged. Maybe understanding this will give me a better grasp of the theory at hand.
Meathands
Well-known member
monkeyxx said:
SCA uses a stepped attenuator and then a trim pot, which is dandy for his modular format but for this kind of preamp I would skip it and use a single reverse log 10k pot for volume control. If cheap and simple are included in your design goals, I would stick closer to the circuit recommended by THAT (see attached). For variable gain control, simply replace RG with a reverse log 10k pot wired as a rheostat (variable resistor).
Since you are a self-proclaimed beginner and you're going to be prototyping this thing without a PCB, I would recommend starting with the basic circuit in Fig 4 of the THAT1512 datasheet without phantom power. I built four channels like this and four with phantom for my rack.
monkeyxx said:
Someone smarter than I will have to answer your first question. Maybe the input impedance of the 2520 itself determines the input Z? The line out impedance is determined simply by the source impedance of the 2520 transformed by whatever turns ratio is used for the 2503.
Attachments
Well, I am a beginner to electronics theory, but I am quite experienced as a builder. I want to learn more theory, it's what I'm lacking. Yes I wasn't going to use the stepped gain switch, I had planned to opt for a pot instead. I just think I like the tweaks that were made to the T15 circuit because I owned two of them and really loved the sound. However the Dantimax preamps on your website sound example section do sound pretty nice. I think SCA managed to get an extra 10 db of gain out of the circuit as well, which I view as a plus. I also like the addition of a full fader trim pot, for padding.
My question at this point is sort of a practical one. It needs to be answered before I can build. I'm trying to figure out if the OPA134PA will be happy driving an output transformer, or if it's going to need a load resistor or not and how to find out the necessary value. I looked at the OPA134PA datasheet but couldn't find anything like "minimum load." If I leave the transformer unterminated, the impedance would be set by the the next stage, an ECHO Audiofire 12 with an input impedance of 10K ohms, which I assume would be fine. I am wondering about the application of an output load resistor on a microphone preamp transformer since they always seem to be driving line inputs of high impedance.
My question at this point is sort of a practical one. It needs to be answered before I can build. I'm trying to figure out if the OPA134PA will be happy driving an output transformer, or if it's going to need a load resistor or not and how to find out the necessary value. I looked at the OPA134PA datasheet but couldn't find anything like "minimum load." If I leave the transformer unterminated, the impedance would be set by the the next stage, an ECHO Audiofire 12 with an input impedance of 10K ohms, which I assume would be fine. I am wondering about the application of an output load resistor on a microphone preamp transformer since they always seem to be driving line inputs of high impedance.
Meathands
Well-known member
No offense intended. I'm a beginner, too, by any standard.monkeyxx said:
monkeyxx said:
The gain stages in the T15 are the THAT 1512 and 1646. The 1646 does add 6dB of gain--where are you seing the extra 10dB? If you do want more gain without the 1646, you can increase the turns ratio of the output transformer or configure the output buffer to provide voltage gain.
monkeyxx said:
What are you wanting to pad? My understanding is that the 10k pot in the T15 schematic provides for finer gain adjustments after the gain switch. This would be redundant if you use a pot.
monkeyxx said:
Cool, since you're interested in the theory this is a great place to apply Ohm's law. Let's say you want your opamp to be able to deliver a massive +28dB of signal into your Audiofire. That's roughly 20V. With a 1:1 output xfmr, the opamp sees a 10k load impedance. 20/10,000 = 2mA is the amount of current that opamp will need to drive the transformer. According to the datasheet, the OPA134 can provide 35mA so you're fine. Or, looking at it another way, the closed loop (configured as a buffer) output impedance of the OPA134 is 0.01 Ohms. It could drive all 12 inputs of the Audiofire and be fine.
monkeyxx said:
This is usually done when you're trying to get color out of an opamp by making it work harder than it wants to. It can also be used to damp the ringing of the output transformer or tune the resonant frequency. A 10k line input will give some transformers lots of room to ring if you don't load them down.
Thanks! I have some huge holes in my theory, and I haven't applied it much yet, but these pointers are very helpful. I think that pretty much answers all of my questions at this point, too, the practical ones anyway. (There's some studying I need to do.) Fixing transformer ringing, I just read, seems to require an oscilloscope, which I don't have. Hopefully the transformers I'm looking at are high quality enough not to have ringing problems. I think most transformers with specific zobel network requirements will list them in the datasheet. I will need to measure and trim any DC offset from the OPA134PA as to not destroy an output transformer when used.
I noticed most of the SCA preamps have an optional 604 ohm load resistor you can add to the secondary of the output transformers, he says this "improves frequency response." If I use the turns ratio squared formula (correctly?) it seems like 600 ohms on the secondary makes the load on the primary see 150 ohms, assuming a 1:2 wiring of the transformer. That's 2^2 x Z = 600, so Z = 600 / 4, which is 150 ohms. However this is also, in the N72 schematic anyway, also occurring at the same time as a Zobel network, in parallel. How do these loads combine? Maybe I should read about series and parallel circuits? You said earlier that the resistive load a transformer is terminated to is what the circuit sees instead of what comes after it, such as a line input, when left unterminated?
If you do the same calculation with a 10K line input as the load on the seconardy, the primary would see 2,500 ohms. Which is above the 1512's minimum stated load assuming it's directly coupled to the transformer. I guess a 1:1 turns ratio would show the 1512 a full 10,000 ohm load. Am I getting the hang of this? Who knows haha!
This discussion has prompted me to take a free online course on transformers and their turns, voltage, current, and impedance ratios. The impedance ratio formula looked especially familiar. I'm actually solving some math problems right now. I had to brush up on my disused algebra skills!
I noticed most of the SCA preamps have an optional 604 ohm load resistor you can add to the secondary of the output transformers, he says this "improves frequency response." If I use the turns ratio squared formula (correctly?) it seems like 600 ohms on the secondary makes the load on the primary see 150 ohms, assuming a 1:2 wiring of the transformer. That's 2^2 x Z = 600, so Z = 600 / 4, which is 150 ohms. However this is also, in the N72 schematic anyway, also occurring at the same time as a Zobel network, in parallel. How do these loads combine? Maybe I should read about series and parallel circuits? You said earlier that the resistive load a transformer is terminated to is what the circuit sees instead of what comes after it, such as a line input, when left unterminated?
If you do the same calculation with a 10K line input as the load on the seconardy, the primary would see 2,500 ohms. Which is above the 1512's minimum stated load assuming it's directly coupled to the transformer. I guess a 1:1 turns ratio would show the 1512 a full 10,000 ohm load. Am I getting the hang of this? Who knows haha!
This discussion has prompted me to take a free online course on transformers and their turns, voltage, current, and impedance ratios. The impedance ratio formula looked especially familiar. I'm actually solving some math problems right now. I had to brush up on my disused algebra skills!
Meathands
Well-known member
Yup, look at the formula for parallel resistance. When you terminate an output tx you are adding impedance in parallel with the input z of the next device. If the terminating R is significantly lower than the following input Z it dominates and essentially sets the load seen by the tx. I would guess that the optional 604R flattens the response by damping the tx.monkeyxx said:
Booya.monkeyxx said:
Link please!monkeyxx said:
Andy Peters
Well-known member
monkeyxx said:
I'm not sure if when you say "minimum load" you mean "minimum load impedance" or not. Because a "minimal load" would actually be a high impedance. "Maximum load" really means "minimum load impedance." Weird, eh?
Anyways, most op-amps really don't need a minimum load and if you remember your op-amp theory, you'll realize that even if you don't put a load on the op-amp, the feedback resistor is also a load. This also means that the feedback resistor is in parallel with any load you hang off the output.
-a
Thanks for the inspiration and elucidation, it helps a lot.
http://www.wisc-online.com/ListObjects.aspx if you click this link then on the left menu "Technical" then after that "Electronics - AC" then there's a long list of items and on the list is "Transformer Practice Problems" where you can learn and practice the turns ratio, impedance ratio, voltage ratio, and current ratio formulas, as well as a brief mention of power always being equal on both sides of a transformer. As you see there's a lot of other topics on this site, I think I'll be spending some time with a few of them in addition to my other study materials. I especially like the practice problem ones.
Meathands said:
http://www.wisc-online.com/ListObjects.aspx if you click this link then on the left menu "Technical" then after that "Electronics - AC" then there's a long list of items and on the list is "Transformer Practice Problems" where you can learn and practice the turns ratio, impedance ratio, voltage ratio, and current ratio formulas, as well as a brief mention of power always being equal on both sides of a transformer. As you see there's a lot of other topics on this site, I think I'll be spending some time with a few of them in addition to my other study materials. I especially like the practice problem ones.
