EEVblog Electronics Community Forum
Electronics => RF, Microwave, Ham Radio => Topic started by: paul002 on May 29, 2020, 08:19:56 am
-
I was building some bandpass filters but having troubles with the 20m (14 mc) ones. I was using T68-6 iron-powder cores which should operate at that frequency. After measuring the Q of the coils it looked that the iron powder cores don't match up to the specs. They are from aliexpress in china. I am curious if somebody has measured real amidon T68-6 cores. I think I have to buy some real cores.
To measure the Q I use the same methode as for measuring crystals. That is have 2 tranformes to create low Z (12,5 ohm) have a series resonat tank circuit with well known C (mica) and use the nanovna to measure the loss in db and calculate the ESR. This can than be transformed to Q = Zl/ESR
-
What Q did you expect, and what Q did you measured?
-
I was expecting something between 200 and 240. But I was measuring only 6 for a chineese T68-6 core on 14 Mhz. I repeated the same measurements at lower frequencies there it would go up to 100. Measuring a air coil at 14 Mhz gave me 651.
-
can you share S1P file with core measurement? And details about coil and capacitor used for measurement
According to your pictures for 16 MHz, Q is about 3.
-
To measure the Q I use the same methode as for measuring crystals. That is have 2 tranformes to create low Z (12,5 ohm) have a series resonat tank circuit with well known C (mica) and use the nanovna to measure the loss in db and calculate the ESR. This can than be transformed to Q = Zl/ESR
I think you got bad Q due to loss in your transformers. Just try to measure it with no transformers
-
Just measured an other one, this one is about 200. Looks much better. I don't think the transformers are an issue. I used them for a lot of measurements and they come from a high quality source. I also measured the Q with the scope using a square wave of 1 Khz and then measure how fast the signal drops to 0.707 V if you multiply by 4.53 you got the loaded Q.
The coils is 0.3862 uH (6 turns 1mm wire T68-1) Cap 248,2 pf loaded Q 1.54 Unloaded Q is Zl/ESR = 38,7 / 0,19 = 200.
If I short the measure points the loss is 0.2 db the loss of the series tank is 0.263 db -> ESR of 0,19 Ohm
-
Measured an other one. Here the ESR is 0,71 ohm and XL 40,35 give a q of 56 so it goes up and down a bit.
I don't have the first one any more so cannot replicate it. But 56 is also quite low for a T68-6.
Will order some from germany see how these are.
-
Just measured an other one, this one is about 200. Looks much better. I don't think the transformers are an issue.
I don't understand for what reason you're using transformer. Because transformer has losses and decrease Q.
For example, just tested Chinese T50-6 with NanoVNA. There is 22 turns with copper wire from network UTP cable. Directly connected to NanoVNA.
It shows Q about 115-120 at 14 MHz. But note, there is no reliable contact, I just put two wires into BNC connector. I think with reliable contact Q will be a little higher.
-
Measured an other one. Here the ESR is 0,71 ohm and XL 40,35 give a q of 56 so it goes up and down a bit.
Your circuit has:
BW = 5.2794 / 15.91703 MHz = 0.33168
Q = 1 / 0.33168 = 3.015
Try to measure coil directly with no transformers. And use more turns to increase wire resistance VNA has worse precision when resistance is too low or too high
-
Just tried without transformers. Does not work. The nanovna does not get -3db points. So cannot measure the q.
What software for the nanovna do you use? I don't see this chart in nano-saver.
The point was that bandpass filters with these cores were having high insertion loss. Trying to measure the Q with the scope also indicated low Q.
When measuring a series resonance you will get low Z, so I used the setup to get a low Z (50 Ohm -> 12,5 Ohm) this setup I also use to measure crystals which works quite good.
Anyway, have ordered some amidon cores and will measure them to see if that is any better or I am just measuring the wrong way.
-
Just tried without transformers. Does not work. The nanovna does not get -3db points. So cannot measure the q.
Can you give S1P file with coil of 20-25 turns on your T68-6 directly connected to VNA port?
-
here it is. sorry is 6 turns, will do a new one
Add a background artikel I was using. But did not try the swr method yet.
-
again now 20 turns
-
paul002, something strange with your core. Are you sure, this is T68-6?
With 20 turns T68-6 should be about L=2 uH, but your data shows L=5.06 uH and Q=16.36 at 14 MHz.
-
yes it is T68-6. Or maybe somebody in china painted it that way >:(.
But it is a good tip I have from 2 suppliers will check them with 20 turns and see if they all are like this.
I also wondered all the time why the count of coil64 was not correct but now I know.
Thanks !
-
it looks like remarked low frequency core. For example 100 kHz or something like that.
-
That is possible. I measured the coil on the scope and it gives me 2,4 uH and a low q about 50. Any way. I wait for the new one and make a comparison. I used them in bandpass filter for 40m and 80m there they work reasonable few db insertion loss, but if the frequency goes up the q drops quickly. Thank for your help
-
Hi ya, I hope you find this helpful, this is 19 turns on a T68-6 genuine I miscounted the actual windings and did not have enough wire left for 20 turns LOL
[attach=1]
[attach=2]
[attach=3]
[attach=4]
-
Can you Measure the q of the cool at 14 MHz? A way to do that is to Measure series resonance and get the -3db points. Using analisys button and choose bandpass filter
-
Can you Measure the q of the cool at 14 MHz? A way to do that is to Measure series resonance and get the -3db points. Using analisys button and choose bandpass filter
3db of what, the plot is in Z not dB
-
this is 19 turns on a T68-6 genuine I miscounted the actual windings and did not have enough wire left for 20 turns LOL
can you show Rs/Xs chart from 0.1 to 30 MHz?
-
this is 19 turns on a T68-6 genuine I miscounted the actual windings and did not have enough wire left for 20 turns LOL
can you show Rs/Xs chart from 0.1 to 30 MHz?
There is no such option, there is R+jX or R/w @ X/w
-
There is no such option, there is R+jX or R/w @ X/w
R+jX chart from 0.1 to 30 MHz
-
There is no such option, there is R+jX or R/w @ X/w
R+jX chart from 0.1 to 30 MHz
Ok whats the math I need to work out Q from here?
[attachimg=1]
-
Q = |X| / Rs or Rp / |X| see article of Jacques Audet, VE2AZX
-
Q = |X| / Rs or Rp / |X| see article of Jacques Audet, VE2AZX
Yeah I think there is something wrong with my first plot. I will re-calibrate the vna tomorrow morning and make up a new test jig. 1
EDIT: did the calibration and got out a quality hardline patch lead and this is what i got. Q of about 150 at 14mhz. Will redo it in the morning when I am less tired and make a new test jig to eliminate any other error. A value somewhere between 50 and 200 would be typically expected. And this value is within that range.
[attachimg=1]
-
Just received new cores, but to my supprise they are also fake >:(. Unbelievable will return it. I did not imagine that german company sell stuff like this. 20 trurns give 2.4 uH instead of 2.0 that already made me suspicious. Now to find a good source of cores.
-
A long time ago Micrometals published a warning to it's customers saying that Chinese equivalents were not the same. Buyer beware.
-
Just received new cores from Reichelt, but to my supprise they are also fake >:(. Unbelievable will return it. I did not imagine that german company sell stuff like this. 20 trurns give 2.4 uH instead of 2.0 that already made me suspicious. Now to find a good source of cores.
Diz from kits and parts in the usa or mark from minikits in australia
-
Thanks will go there.
Just received new cores from Reichelt, but to my supprise they are also fake >:(. Unbelievable will return it. I did not imagine that german company sell stuff like this. 20 trurns give 2.4 uH instead of 2.0 that already made me suspicious. Now to find a good source of cores.
Diz from kits and parts in the usa or mark from minikits in australia
-
It looks like ow mr Reichelt an apology :palm: . Somehow I was not happy with the measurement of the cores I received yesterday. Did some research and found a methode described by W7ZOI in his evaluation of the nanovna. He describes that measuring high Q will not always work on a (nano)vna. So I tried that methode. You create a pass thrue connector with two sma female connectors and connect the core in serie resonance to ground. Measure the S21 gain and calculate the q described in the book Experimental Methodes in RF design. Now the Chinese core still measures a Q of 31 and the new core from Reichelt a Q of 200.
Somehow the methode of connecting the core direct to the S11 port does not work for me.