Sometimes the transformer inbuilt temperature fuse (120°C) burned. And the whole transformer had to be replaced. Prefer using tansformers with ambient temperature range TE40 and not TE60. Or not to use a transformer to 100% of its capability. Some more reserve seems to be prevent overheating especially on summertime.
By the way: Most of the DC-stages do not strictly follow the datasheet recommendations for mounting rectifier diodes at least 9.5 mm lead length to prevent overheating. You may see the result of such mounting to tough to the printed board in changing the color to "charcoale like". That is why the rectifier in my very oldschool and a little bit oversized PSU is a category 25 Amps one and of course mounted on metal case for cooling.
And a transformer can reach 115 to 120° Centigrade inside the coil, quoting Category E. Even if it is a class T40 one. In such most cases a temperature fuse has been fitted inside the primary coil set for 120°C, which cannot be replaced so easily.
In my linear supplies, I had to swap conventional silicon rectifiers for Schottky types, as the 1.6 voltage drop and 3A + forward current was producing about 5 w of heat.
The downside of Schottky types, is that they have faster switching times, and as a result, they can produce additional RF interference, unless extra R & C snubbing components are added.
As you know, not all kind of so called "Schottky"-diodes are best for that mentioned rectifier purposes. Some developments have had took place in the meantime. SiC (Silicium Carbid) Schottky seeming to be suitable, despite a look on diagrams show dependency of forward and backward current in correlation to current and temperature, and that look on might not be totally overwhelming. And those diodes got package for heatsink mountig, already. Needing a kind of heatsink anyway.
As G8JNJ mentioned above, here a very remarkable experiene, what could happen, if snubber has not been calculated out properly. Approximately half a tenthousand (or more - I do not know) of powersupplies of very recommended popular manufacturer were called back due to possible risk of fire. The schottky diode probably in use was SR3100 and needed a resistor of 10 Ohms in series to a condensor of 3.3 nF. Mainly the resistor overheated. That snubber was no more necessary, when using centere tapped secondary and a double diode such as SF163A (common anode) or SF163C (for common catode) mounted on a small heatsink.
I am planning on purchasing a KiwiSDR but just learned of the SMPS "dirty" power supply issue. I fear I do not have the skills to build a linear supply and the KiwiSDR folks have one they have tested for $90.00 Do you know if any cheaper but acceptable options? Also, I guess I should put Ferrite RF chokes on the wires going into and out of the SDR.. ??? I have an enterprise Ubiquity lan/Wifi network in my home , I guess I should but the SDR on a long LAN run away from all that gear?
Unfortunately there are some fakes on the road, which disguise themselves in TO-3 housing, but never deliver the higher load current than approximately 1A. However you can easily sort them out by having a look on the mounting side. Some butterfly-like chip welding got the original ICs. This characteristic construction sign is valid also for other voltages e.g. 12V ICs as I could recognize.
I have three SDR receivers and a linear power supply for three channels of 5 volts. I'm using an LT1084ADJ in a TO220 package. It has been working around the clock for over a year now. And I think the LT1083/1084/1085 stabilizer is a very good and cheap choice.
Martin, thank you. I'll think about replacing my LT1084 with LM338 . No changes to the adj schematic are required. They have the same reference voltage.
The maior problem is the availability of original parts. Got a LM338T in a TO-220 housing. This part is heating up too much, even proper mounting on heat sink. The housing TO-3 should be preferred used. However me did not find any trusted vendor. The parts from outside the EU did not match the database requirements. E.G. There are inside chips for lower current. And the US manufacturer quote high postage costs approx. 40 $ per delivery. This is not worth while. I my consideration, will get back to old fashioned mains supply with 2N3055 as power regulator transistor. The higher ripple on output DC does not matter.
Great discussion. I hope this isn't too far off-topic, but I've been trying to find an extremely low-noise 19 VDC PSU (at 5 to 6 A) to power an Intel NUC when used with SDRs at frequencies below 500 kHz.
I bought a Super Hiend 19V 5A Linear Power Supply For Intel NUC from fyfaudio.com but am finding the noise at in that frequency range is non-trivial.
Any suggestions on adapting ideas from this thread for that voltage/amperage? I don't mind going with a couple of automotive-type lead acid batteries if I could find a regulator that would drop their combined 24v to 19v without adding noise.
I recently did a bit more research into the various DC filter boards that can be bough cheaply from China.
I've previously mentioned the simple type with two separate inductors.
Simple DC Filter
But I wanted to compare it against some of the larger types with an additional common mode choke using a dual winding. All of these seem to use the same printed circuit board, but populated with different types of inductors.
Green rectangular core common mode choke, and black core inductors
With toroidal common mode choke and yellow / white core inductors (Type 26 Iron Powder)
Differential Mode attenuation - direct noise from the power rails
All the filters are pretty effective. The simple type is slightly worse, but I doubt you would notice much difference between them in "real life"
Common mode attenuation - conducted noise, usually via a ground path
The Green and White cored filters are much more effective. The White cored filter would be slightly better at reducing low frequency <1MHz switching noise from Switched Mode Power Supplies, but the Green cored version provides much better overall attenuation.
Note that the values of attenuation were measured in a 50 ohm system, and the actual amount of attenuation will depend on the circuit impedance. However, using 50 ohms does at least provide a standard reference for comparison purposes.
Comments
The spectrum display is much more useful in measuring RFI than the waterfall
I built a nice linear power supply.
Some parts were new, some old.
It died this morning. :-(
It will have to wait until the weekend.
Back to the Apple wall warts.
Well, finally had time to fix it. Turns out it was a fuse inline with the AC input. :-D
Sometimes the transformer inbuilt temperature fuse (120°C) burned. And the whole transformer had to be replaced. Prefer using tansformers with ambient temperature range TE40 and not TE60. Or not to use a transformer to 100% of its capability. Some more reserve seems to be prevent overheating especially on summertime.
The output is 13.8V 1.5A, and the DC output is 5V 2.5A.
The transformer does run warm, however its a comfortable temp.
The DC stage is one of these.
For the price, I couldn't buy the parts locally cheaper.
By the way: Most of the DC-stages do not strictly follow the datasheet recommendations for mounting rectifier diodes at least 9.5 mm lead length to prevent overheating. You may see the result of such mounting to tough to the printed board in changing the color to "charcoale like". That is why the rectifier in my very oldschool and a little bit oversized PSU is a category 25 Amps one and of course mounted on metal case for cooling.
And a transformer can reach 115 to 120° Centigrade inside the coil, quoting Category E. Even if it is a class T40 one. In such most cases a temperature fuse has been fitted inside the primary coil set for 120°C, which cannot be replaced so easily.
In my linear supplies, I had to swap conventional silicon rectifiers for Schottky types, as the 1.6 voltage drop and 3A + forward current was producing about 5 w of heat.
The downside of Schottky types, is that they have faster switching times, and as a result, they can produce additional RF interference, unless extra R & C snubbing components are added.
Regards,
Martin
As you know, not all kind of so called "Schottky"-diodes are best for that mentioned rectifier purposes. Some developments have had took place in the meantime. SiC (Silicium Carbid) Schottky seeming to be suitable, despite a look on diagrams show dependency of forward and backward current in correlation to current and temperature, and that look on might not be totally overwhelming. And those diodes got package for heatsink mountig, already. Needing a kind of heatsink anyway.
As G8JNJ mentioned above, here a very remarkable experiene, what could happen, if snubber has not been calculated out properly. Approximately half a tenthousand (or more - I do not know) of powersupplies of very recommended popular manufacturer were called back due to possible risk of fire. The schottky diode probably in use was SR3100 and needed a resistor of 10 Ohms in series to a condensor of 3.3 nF. Mainly the resistor overheated. That snubber was no more necessary, when using centere tapped secondary and a double diode such as SF163A (common anode) or SF163C (for common catode) mounted on a small heatsink.
SIC_SF163A_C_w/o_snubber.jpg first picture below
SIC_diode_snubber_.jpg
Hello :)
I am planning on purchasing a KiwiSDR but just learned of the SMPS "dirty" power supply issue. I fear I do not have the skills to build a linear supply and the KiwiSDR folks have one they have tested for $90.00 Do you know if any cheaper but acceptable options? Also, I guess I should put Ferrite RF chokes on the wires going into and out of the SDR.. ??? I have an enterprise Ubiquity lan/Wifi network in my home , I guess I should but the SDR on a long LAN run away from all that gear?
Thanks for any help 😀
I did a bit more homework and found my questions already answered elsewhere on the Forum.
Thanks to the Super Users who posted those answers 😀
Beware of Ubiquiti radio links using PoE.
The DC-DC convertors and plastic housings, can make them very noisy on HF frequencies.
Regards,
Martin
Now working fine 24/7 with an 5V-IC featuring maximum 5 A in a TO-3 housing.
https://forum.kiwisdr.com/uploads/430/MDIR0FHIJD6E.jpg
Unfortunately there are some fakes on the road, which disguise themselves in TO-3 housing, but never deliver the higher load current than approximately 1A. However you can easily sort them out by having a look on the mounting side. Some butterfly-like chip welding got the original ICs. This characteristic construction sign is valid also for other voltages e.g. 12V ICs as I could recognize.
Are you sure an LM317 is the right choice? The current that the Kiwi needs is a constant 1A and reaches 1.5A+
you might want to look at an LM350 design.
I purchased one of these off ebay
https://www.ebay.com/itm/145022571057
It runs quite hot, however it works well.
I have three SDR receivers and a linear power supply for three channels of 5 volts. I'm using an LT1084ADJ in a TO220 package. It has been working around the clock for over a year now. And I think the LT1083/1084/1085 stabilizer is a very good and cheap choice.
"And I think the LT1083/1084/1085 stabilizer is a very good and cheap choice."
The 5A linear regulators do produce LF noise.
But adding a big electrolytic across the regulated output helps a lot.
Regards,
Martin
Martin, thank you. I'll think about replacing my LT1084 with LM338 . No changes to the adj schematic are required. They have the same reference voltage.
The maior problem is the availability of original parts. Got a LM338T in a TO-220 housing. This part is heating up too much, even proper mounting on heat sink. The housing TO-3 should be preferred used. However me did not find any trusted vendor. The parts from outside the EU did not match the database requirements. E.G. There are inside chips for lower current. And the US manufacturer quote high postage costs approx. 40 $ per delivery. This is not worth while. I my consideration, will get back to old fashioned mains supply with 2N3055 as power regulator transistor. The higher ripple on output DC does not matter.
Great discussion. I hope this isn't too far off-topic, but I've been trying to find an extremely low-noise 19 VDC PSU (at 5 to 6 A) to power an Intel NUC when used with SDRs at frequencies below 500 kHz.
I bought a Super Hiend 19V 5A Linear Power Supply For Intel NUC from fyfaudio.com but am finding the noise at in that frequency range is non-trivial.
Any suggestions on adapting ideas from this thread for that voltage/amperage? I don't mind going with a couple of automotive-type lead acid batteries if I could find a regulator that would drop their combined 24v to 19v without adding noise.
I recently did a bit more research into the various DC filter boards that can be bough cheaply from China.
I've previously mentioned the simple type with two separate inductors.
Simple DC Filter
But I wanted to compare it against some of the larger types with an additional common mode choke using a dual winding. All of these seem to use the same printed circuit board, but populated with different types of inductors.
Green rectangular core common mode choke, and black core inductors
With toroidal common mode choke and yellow / white core inductors (Type 26 Iron Powder)
Differential Mode attenuation - direct noise from the power rails
All the filters are pretty effective. The simple type is slightly worse, but I doubt you would notice much difference between them in "real life"
Common mode attenuation - conducted noise, usually via a ground path
The Green and White cored filters are much more effective. The White cored filter would be slightly better at reducing low frequency <1MHz switching noise from Switched Mode Power Supplies, but the Green cored version provides much better overall attenuation.
Note that the values of attenuation were measured in a 50 ohm system, and the actual amount of attenuation will depend on the circuit impedance. However, using 50 ohms does at least provide a standard reference for comparison purposes.
Regards,
Martin
Hi Martin
Interesting analysis.
Have you tried testing whilst DC was flowing to see if saturation of the core happens and if it reduces the effectiveness of the filters.
Judging by the photo, the simple filter probably can't carry as much DC current as the other two.
kind regards
Steve G8KDL