Thanks for your detailed explanation.
I am aware that all these factors are very relevant. I just want something that approximates a 1MHz to 10MHz bandwidth commercial amp, but without the price tag.
I wouldn't expect lab performance, and know this involves compromises in fidelity, and no extra features.
No voltage gain. Fixed frequency. No sweep. No offset. AC or DC coupled. Anything that gives the impression of working.
Barry Latham
if all you want is fixed-frequency sine waves with some distortion, the emitter followers should do just fine.
Interesting part (its kinda like the TCA0372) but the slew rate is insufficient to go far beyond 1MHz. IMHO you can use this chip in DC coupled applications as well.
Nice, I was trolling digikey for high current opamps the other day and found the LT1206. It's a little light on the current side... only 250mA, but...
George H.
Only 1 MHz? The data sheet says 900v/usec slew rate. For a 20v swing, that's: Slew_Rate = 2*Pi * Freq * Voltage_swing or Freq = Slew_Rate / ( 2*Pi * Voltage_swing) Freq = 900V/usec / ( 6.28 * 20V) = Freq = 900*10^6V/sec / (6.28 * 20V) Freq = 7.2 MHz. A 10Mhz sine wave will work, but a 10 MHz square wave would be a problem.
Sure. Don't forget the heat sink. See the Thermal Considerations in the data sheet: on Pg 11.
Sigh... ok.
It will be a multi-winding torroid, wound in a bifilar manner. Each LT1210 has a 10 ohm output impedance. Two in series, as in the bridge amp, make 20 ohms. To get to 50 ohm requires an impedance ratio of: 50 / 20 = 2.5 or a turns ratio of: 2.5^0.5 = 1.5:1 or converted to integer number of windings: 1.5:1 = 2:3 turns ratio To do that requires 5 windings on the torroid. The numbers 1 - 5 below represent the ENDS of the 5 wires: 1-------------------\ 2----------1 3 3----------2 turns /------4 3--------/ 2 turns 5----------4 \-----------------5
All 5 wires should be twisted together to provide maximum coupling at the higher frequency end.
I'll cheat and reverse engineer the Coilcraft part. Each of the six windings have an inductance of 22.3uH. The two turn primary winding has a reactance of: Xl = 2*Pi*F*L = 6.28 * 100*10^3 * 22.3*10-6 = 14 ohms @ 100 Khz The two windings are in series for a total of: 14 * 2^2 = 56 ohms and 14 * 3^2 = 126 ohms Those are high enough reactance so that there's little gain loss with Xl=56 ohms across the 10 ohm LT1210 output impedance.
Next comes the torroid size, material selection, and wire gauge: I would like to do this with a reasonable number of turns. Offhand, I'll start with 5 turns which requires: Al = 22.3 uH / 5t = 446 uH/100t Using type 43 material (and I like coated torroids): shows an Al = 480, which is close enough. The 0.312" hole might be a bit tight for 5 wires and 5 turns but I think it will fit. Let's check if it will: Inside circumference = 0.312" * Pi = 0.98 inches 0.98in / 25 wire_diameters = 0.039 in/wire Looking at the wire table at: Anything smaller than #20 AWG will fit neatly. Peak current per LT1210 is 2A which #20 AWG can easily handle without turning into a fuse. However, I would use something smaller like #24 or #26 to make the windings easier to fit.
Good luck keeping the wires organized. This might help:
Note that the 1000 pF capacitor C4 in the LTspice model will change. It resonates with the transformer primary at about 20 MHz to produce a high end gain boost. You can see the peak in the freq response curve:
Not the most elegant design method, but this should be sufficient to put something together that works.
I picked Amidon because it's commonly available. Since you're not going into production with this contrivance, I saw no reason to find a cheaper source.
Reminder. The design is not DC coupled, which might have undesireable results for whatever you're doing.
FWIW, I think the Coilcraft thing was wound in a hexafilar manner.
"Jeff Liebermann"
** The rise time and/or slew rate of a "square wave" is undefined.Only fools talk as if it is.
.... Phil
Guilty as charged. I called it "wound in a bifilar manner" so that a Google search of "bifilar wire" would find something useful. That was intentional as I suspected that "pentafilar" would not return anything useful. I guess "multifilar" would have been a better choice for winding terminology.
Incidentally, quadrifilar and hexafilar return URLs for circularly polarized GPS antennas.
118,000 hits, and:
33,500 hits, which suggests that there are quite a few fools out there.
Come on Phil, there ain't no 'ideal' square waves. (You'd need frequencies out to infinity... x-rays!) They've all got rise times/ slew rates.
George H.
"Jeff Liebermann" "Phil Allison"
** There are literally millions of them.But very few are as big a f*****ad as you.
I said: " The rise time and/or slew rate of a "square wave" is undefined."
It is of course possible to define those parameters with a particular example, but no general definition exists.
.... Phil
** There are only non ideal ones and there is nothing defining what a " square wave" should be. ** Only real ones do.
Q. What is the RT and SR of a 10MHz square wave ?
A. No answer is possible, because it is undefined.
Capice now - f****it ?
.... Phil
Grin, Phil you give me a chuckle. (please don't get too upset.) I only have to deal with 'real' square waves. If you want treat square waves, like theoretical delta functions, we can agree to disagree.
It's late for me over here, I've got a half a glass of milk left before bed. Cheers, George H.
I also get relevant hits like this one:
"Suitable off-the-shelf components are available, such as the Coiltronics Versa-Pac? series. These are hexafilar wound and give power bandwidths in excess of 10MHz. One disadvantage is that using a limited number of 1:1 windings makes it impossible to exactly transform 50W to the optimum 10W load. Nevertheless, there are several useful connections."
..and this interesting lower-frequency offering from Coilcraft.
And some coil winding data from Power Integrations for SMPS designs.
Best regards, Spehro Pefhany
That's the app note I referenced earlier in this thread for the LT1210 bridge amp design (Fig 70). Incidentally, when I type in just "hexafilar" in the Google search box, the above app note doesn't show up in the first 100 hits that I checked. Did you use a different search engine or method?
Good find. US version: I dunno about the "Frequency range up to 1 MHz".
The Coilcraft inductors look like either bobbin wound or E-I stack inductors, not a torroid. The original app note specified a VPH5-0155R. The above data sheet shows an HPH5-0155L which has the exact same inductance (22.3uH) and required saturation current (2A). So, the HPH5-0155L is available, it should work as a suitable substitute.
In the old days (15 years ago) a Denon amplifier (PMA1080 IIRC) would have a rarther large bandwidth. AFAIR over 1MHz. I have used it for sub mohm impedance measurements
Can't beat the price, find one at CraigList (or whatever it is called in the states), probably around 50 USD
Cheers
Klaus
Le Wed, 02 Oct 2013 00:45:44 -0400, Spehro Pefhany a écrit:
Nice!
But I just hope that, for the sake of datasheet accuracy, they rounded the MTBF minutes and seconds to the nearest hour:
Mean Time Between Failures (MTBF)
26,315,789 hours
Your units don't work out right: how do you get more-per-100t by dividing by a smaller number?
A_L is actually per turns squared, usually uH or nH. So 22.3uH in 5t should be 0.892 uH/t^2. There is no such thing at "uH/100t", only laziness -- what's meant is "uH/(100t)^2".
Ferrites are usually nH or uH/t^2, and the unit is specified (except at lazy distributors, like Amidon...sigh). RF cores (mainly from oooold catalogs) are sometimes in uH or mH at 100t, which isn't off by a power-of-1000 multiplier. Read carefully, and if possible, find the manufacturer datasheets! (#43 ferrite is a Fair-Rite designation. Amidon doesn't specify this for some reason.)
Anyways, it looks like the core you've found is a bit marginal -- you could add more turns, or stack two, or you might consider ferrite beads instead, which are generally inefficient for transformer use, but have certain advantages in applications like this, due to their high inductivity.
I suppose another advantage to ferrite beads: everyone carries them. Though Amidon's price on them is reasonable, so either way is good.
Tim
You're being pedantic. All my trapezoidal waves have a risetime spec. I define it. I look in the equipment catalog and compare the vendor's definition to mine. Purchase decision is made considering that comparison.
In the analog oscilloscope days there were some generally accepted rules of thumb. 50 MHz. got you 7ns risetime. And there's math behind that based on a compromise between transient response and frequency response. High-order digital processing can mess with those numbers, but they're still reasonable for assessing analog function generator performance.
Back in the day, I compared some commercial function generators. The maximum frequency of the function generator appeared to be the frequency at which you couldn't tell the difference between sine, square and triangle. And the waveform wasn't a particularly accurate representation of any of the three.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required