You won't be able to measure trace inductances with that. I guess I shouldn't have expected 50R output resistance from a generator that only goes to 1MHz. It won't test your amp at the BW limits either. You should get something that will allow you to test up to the BW of your scope. I am lucky to have a Tektronix FG504.
Why don't you try measuring the ESL of a wire? See how short a wire you can measure. Set the SG to max F and V, and short it across the wire, measuring with Kelvin connections directly to the scope. Use alligator clips if you need to.
If you have the money to spend, getting something like the Tek FG504 is ideal. It has lin and log frequency sweep, adjustable risetime, sine/square, triangle/saw with adjustable asymmetry and a VCO input. It took a very good friend to get that, and it still has some quirks I haven't been able to fix yet. Still, it works well enough for what I need it for.
If you have the money to spend, getting something like the Tek FG504 is ideal. It has lin and log frequency sweep, adjustable risetime, sine/square, triangle/saw with adjustable asymmetry and a VCO input. It took a very good friend to get that, and it still has some quirks I haven't been able to fix yet. Still, it works well enough for what I need it for.
The Tek FG504 looks like a lovely bit of kit. I'm jealous! Maybe a bit too much for me at present.
I think something that can drive up to about 25Mhz with a 50 ohm output impedance is the bare min required. Going to do a load of research on this.
Will give your experiment a go. I don't hold out much hope.
Thank you for helping me out so much. It is very much appreciated!
I think something that can drive up to about 25Mhz with a 50 ohm output impedance is the bare min required. Going to do a load of research on this.
Will give your experiment a go. I don't hold out much hope.
Thank you for helping me out so much. It is very much appreciated!
I tried a lash up experiment. The SG useless. I could measure down to about 5cm. So I need more current and/or higher frequency.
PC based SGs offer some hope of getting something within a reasonable budget assuming a 50R output is sufficient.
For now I will progress the PCBs and do more research.
PC based SGs offer some hope of getting something within a reasonable budget assuming a 50R output is sufficient.
For now I will progress the PCBs and do more research.
Perhaps the AADE LC meter (link) might do some of the things you want? It's meant for RF work so it expects you'll be measuring relatively high quality (hi-Q) inductors and/or capacitors, whose inductance or capacitance is relatively stable vs. frequency. 60 Hz power transformers are not the intended application!
I just used it to measure the inductance of a couple pieces of AWG22 hookup wire, length=9cm and also length=2cm; photos below. Seems reasonable to me.
I just used it to measure the inductance of a couple pieces of AWG22 hookup wire, length=9cm and also length=2cm; photos below. Seems reasonable to me.
Attachments
Last edited:
I haven't used the AADE LC meter to measure inductances below 20 nH so I can't offer any opinions. The meter continuously measures and displays, so the reading drifts around, and every once in a while the drift diverges monotonically. I've decided to trust the measured value seen immediately after zero-cal (for L: short the probes; for C: opencircuit the probes), which I write down in my "lab notebook" a/k/a cocktail napkin for a permanent record. I wish the thing had a sample-and-hold button that freezes the display, but it doesn't.
cocktail napkin... ha ha. I normally write things on scraps of paper which then inevitably get thrown away.
This instrument deserves further consideration. It's a reasonably price too. Need to work out the best solution that's within my budget.
For now, my efforts are going to be focussed on the PCBs. I have reworked the output PCB to use that capacitor you suggested. It's perfect for the job and has the prefect dimensions, short and wide.
This instrument deserves further consideration. It's a reasonably price too. Need to work out the best solution that's within my budget.
For now, my efforts are going to be focussed on the PCBs. I have reworked the output PCB to use that capacitor you suggested. It's perfect for the job and has the prefect dimensions, short and wide.
A side note to Ls: I participated briefly in a study of what the average trace Z was in a multi IC/digital/analod board would be. All math and no test equipment: Using typical pcb material, typ trace widths, typ copper thickness and the total number of pins/traces etc; A statitician looked at all the numbers/data and came up with an average Z.... Guess what? It was 50 Ohms.
This was probably not a coincidence but by design.
Thx-RNMarsh
This was probably not a coincidence but by design.
Thx-RNMarsh
How do you know the accuracy of the indicated result? I mean, it reads something but is it really that number shown? How do you know?
Thx-RNMarsh
And this is the question. You could calculate the inductance of a given length of wire then take a measurement. Or buy precision inductors (if they exist) and measure them.
Even if it has an error then as long as its consistent then you could deal with it.
The figures measured by mark are consistent with the rule of thumb of 1nH per mm. But how accurate is the rule of thumb?
Too many variables... Need more thinking time and research.
Even if it has an error then as long as its consistent then you could deal with it.
The figures measured by mark are consistent with the rule of thumb of 1nH per mm. But how accurate is the rule of thumb?
Too many variables... Need more thinking time and research.
The equipment vendor does provide (this web page) which presents quite a bit of measured data. He measures sixty (!) different calibration-standard inductors, with his instrument and also with three other LC meters. Error results are tabulated.
But of course, that just tells you about the accuracy of his LC meter; how do I know whether my LC meter, on my workbench, is accurate? I don't. If someone wants to send me a precision inductor whose value is known to be 89 nH (or 22nH), I'd be glad to measure it with my LC meter and post a photo. PM me for USmail address.
It is true that textbooks give formulae to calculate the Self-Inductance of a Straight Conductor, which are applicable to the case of a 9cm length of AWG22 wire. I consider (Frederick W. Grover's book) to be authoritative / definitive on the topic.
Much to everyone's surprise, Grover gives a formula for the inductance of a straight wire, in which inductance is not linearly proportional to length. I will repeat that: the inductance of a straight wire is not linearly proportional to its length. Instead, inductance is proportional to { length x ln(length) } , i.e. super-linear growth*. Image attached.
Plugging the numbers for the 9 cm length of AWG22 wire, into Grover's equation (7) on page 35:
*As anyone who took an EE class called "Analysis of Algorithms" can tell you, N ain't the same as NlogN, not by a long shot. Finding the max of an unsorted array is order N. Sorting the array is order NlogN.
But of course, that just tells you about the accuracy of his LC meter; how do I know whether my LC meter, on my workbench, is accurate? I don't. If someone wants to send me a precision inductor whose value is known to be 89 nH (or 22nH), I'd be glad to measure it with my LC meter and post a photo. PM me for USmail address.
It is true that textbooks give formulae to calculate the Self-Inductance of a Straight Conductor, which are applicable to the case of a 9cm length of AWG22 wire. I consider (Frederick W. Grover's book) to be authoritative / definitive on the topic.
Much to everyone's surprise, Grover gives a formula for the inductance of a straight wire, in which inductance is not linearly proportional to length. I will repeat that: the inductance of a straight wire is not linearly proportional to its length. Instead, inductance is proportional to { length x ln(length) } , i.e. super-linear growth*. Image attached.
Plugging the numbers for the 9 cm length of AWG22 wire, into Grover's equation (7) on page 35:
l = 9 cm
rho = 0.0644 cm
L = inductance = (2E-3) x 9 x [ ln(2x9/0.0644) - 0.75 ] = 0.0879 uH = 87.9 nH
rho = 0.0644 cm
L = inductance = (2E-3) x 9 x [ ln(2x9/0.0644) - 0.75 ] = 0.0879 uH = 87.9 nH
*As anyone who took an EE class called "Analysis of Algorithms" can tell you, N ain't the same as NlogN, not by a long shot. Finding the max of an unsorted array is order N. Sorting the array is order NlogN.
Attachments
This inductance meter looks very promising and is the right sort of price.
Update on the project so far. There have been some changes to the schematic. A cascode has crept in (Manso's suggestion). Allowing the use more suitable input devices with better matching of operating conditions.
Changed to BC5xx devices where the transistors are subject to low Vce. This was prompted after reading about quasi saturation in the impressive astx thread.
Would like to use BC3xx devices for the output devices but cannot find decent spice models for them. Will be tweaking component values for a while but the basic design has been settled on (for now).
PCBs have also been designed and ordered. So hopefully in the next couple of weeks can start building and experimenting. Time to accumulate the rest of the parts.
Some plots etc for you... if anyone is interested.
Cheers
Paul
Update on the project so far. There have been some changes to the schematic. A cascode has crept in (Manso's suggestion). Allowing the use more suitable input devices with better matching of operating conditions.
Changed to BC5xx devices where the transistors are subject to low Vce. This was prompted after reading about quasi saturation in the impressive astx thread.
Would like to use BC3xx devices for the output devices but cannot find decent spice models for them. Will be tweaking component values for a while but the basic design has been settled on (for now).
PCBs have also been designed and ordered. So hopefully in the next couple of weeks can start building and experimenting. Time to accumulate the rest of the parts.
Some plots etc for you... if anyone is interested.
Cheers
Paul
Attachments
I hope this will not complicate project, but I have to say that from my experience BC550c/560c are bad sounding. Kind of rough sound.
We shall see
. The good thing is I can use the 2n5551/5401 transistors in their place by turning them round 180 degrees.The design doesn't appear to be that sensitive to the different transistors. Just a few resistor changes required.
Fairchild BC5x0C are different from OnSemi BC5x0C. According to datasheets, the OnSemi parts have lower nonlinear capacitance. I'm actually surprised Cordell's models show relatively high Cje for the BC5x0C. By the datasheets, the BC5xx series actually have amazingly low capacitances, and my experience is they do sound rough if they oscillate but are capable of mind-blowing clarity when used very carefully. Anyways, if the Fairchild BC5xxC don't sound good, try the OnSemi parts. I switched them into my EC buffer and I think they sound better.
As far as BC3x7 models, what problems are you encountering? You might try the BC807/817 models from NXP.
As far as BC3x7 models, what problems are you encountering? You might try the BC807/817 models from NXP.
Excuse my inexperience and lack of knowledge in this area. What would cause the BC5xxC to oscillate and what precautions could be taken to prevent this from happening? In spice they seem happy enough, reality may be somewhat different.
The PCB is designed such that there is scope for experimentation. Eg. I realised that since the pin outs for the BC devices were similar to the 2N devices with the base pin in the centre, I could easily swap between the two. Also could loose the input cascode with a few wire links.
I liked the low Cob of the BCs along with the high gain. The BC546 + complement looked very nice with the high gain. They also seem much better for the current sources in spice but this could be down to models.
The problem I'm having with the BC3x7 models is convergence. I can drop the Gmin down to 1e-006 to get anywhere.
Edit: for a laugh simulated 100KHz @ +/-36v ad got 0.004%
The PCB is designed such that there is scope for experimentation. Eg. I realised that since the pin outs for the BC devices were similar to the 2N devices with the base pin in the centre, I could easily swap between the two. Also could loose the input cascode with a few wire links.
I liked the low Cob of the BCs along with the high gain. The BC546 + complement looked very nice with the high gain. They also seem much better for the current sources in spice but this could be down to models.
The problem I'm having with the BC3x7 models is convergence. I can drop the Gmin down to 1e-006 to get anywhere.
Edit: for a laugh simulated 100KHz @ +/-36v ad got 0.004%
Last edited:
I suspect you a right. Have been doing quite a bit of reading about source stepping. It's not quite sunk in what a lot of the settings do. Eg changing Voltol seems to help matters. Just not sure what it relates to.
The funny thing is that the complete parasitics version of the circuit gives a lot less problems with source stepping than the non parasitic version.
The funny thing is that the complete parasitics version of the circuit gives a lot less problems with source stepping than the non parasitic version.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.