Hi guys, A week ago I built a Marshall power brake attenuator from parts I found aroud:https://db.tt/lDdqNREI
The transformer is actually a 100V single-winding speaker matching transformer that I found in my stuff. The inductors are hand wound by myself and the core is from these cheap ferrite rod types:http://www.petervis.com/Radios/making-a-crystal-radio/crystal-radio-coil/ferrite-rod.gif
The rest of parts are as listed in the schematic:http://prowessamplifiers.com/schematics/thumbs/MarshallPower_Brake.gif
The thing works as supposed to, and sounds almost transparent in all attenuation levels, but last day I compared its reactance in combination with a Vintage-30 speaker, to the speaker by itself and plotted the results in Excell: https://db.tt/BgbLPaIO
The impedance is measured with 180mV signal and using a 10R resistor in series with the load, and the attenuator is on the -18dB setting.
Everything in the graph seems as expected except one thing; Why the impedane rises at high frequencies? And how can I defeat it by means of changing component values?
Any help will be appreciated...
QuoteThe impedance is measured with 180mV signal and using a 10R resistor in series with the load, and the attenuator is on the -18dB setting.
i think that when using the power brake at -18dB ther should be a speaker connected to output. if you want to use it without speaker, use position load box
Could you maybe post a drawing of your measurement setup?
Quote from: Vitrolin on February 17, 2016, 12:23:59 PM
i think that when using the power brake at -18dB ther should be a speaker connected to output. if you want to use it without speaker, use position load box
Could you maybe post a drawing of your measurement setup?
Hey man; thank you for the answer. The power brake is actually in combination with the speaker, sorry for not mentioning earlier; I'm not planning to use it as a Load Box. I used my laptop as signal generator and measured the AC volts across the resistor in series with the load in different frequencies; here is my set up:
https://db.tt/eUuHRcJH
That looks like a normal speaker impedance chart to me. After the nominal impedance point (the low point on the graph), the speaker load becomes predominantly inductive which is why the impedance continues to rise at higher frequencies. A better question would be why does the actual speaker seem to taper off at high frequencies? Capacitive loading?
Quote from: PBE6 on February 17, 2016, 03:34:34 PM
That looks like a normal speaker impedance chart to me. After the nominal impedance point (the low point on the graph), the speaker load becomes predominantly inductive which is why the impedance continues to rise at higher frequencies. A better question would be why does the actual speaker seem to taper off at high frequencies? Capacitive loading?
I only measured the impedance to only "5KHz", so if you look at the gragh in the given datasheet, you can see that the speaker line (the blue line on my gragh), is actually alright:
https://guitargear.files.wordpress.com/2009/04/gov_freqresp.jpgOn my test gragh, the resonance peak is limitted by the brake as expected, but I can't see why the impedance rises that much at high frequencies? Could it be the low value cap?
> inductors are hand wound by myself and the core is from these cheap ferrite rod
Ferrite rod for 1MHz radio may not work as expected at 5KHz. How did you verify the values?
The Prowess Amps schematic shows a basic simulated speaker L2 C1 R3 L1.
L2 C1 is the bass resonance. R3 is the voice coil resistance. L1 is the voice coil inductance.
The values are reasonable as far as they go.
I would expect a 50-100 Ohm resistor across L2 C1, but it appears that your coil-loss gives at least this much (an Ohm of winding series resistance will be a lot like 50r of shunt resistance). For guitar amps, you need a bass resonance, but the height of the bass resonance is not very critical.
In a real speaker, the "L1" 1mH part is VERY "lossy". This is the raw voice coil. Which is a coil, but it is surrounded by conductor (iron pole pieces). The more it inducts, the more the iron sucks out the inductance. The actual rise of effective impedance is much less than a pure coil. And less than a coil on ferrite (non-conductor).
The rise of impedance is probably not harmful. While most speakers are poorly inductive, some are more inductive than others.
I would put about 20 Ohms 10 Watts *across* L1. This approximates the effect of the iron around the voice voil. This ensures that impedance will not rise toward infinity. It will actually be a fairly good match to your V30. The exact value of impedance above 3KHz is not too important.
You could also try winding L1 on an iron bolt. The math is un-clear, but a bolt may be cheaper than ferrite rod.
For a super-exact equivalent impedance, L1 would be split in several parts with resistors across each part. However every model of speaker would need a different L-R network to be "exact". Amplifiers are not that fussy. 15 to 30 Ohms across L1 will be good enough for our purposes.
Quote from: PRR on February 18, 2016, 01:48:39 AM
> inductors are hand wound by myself and the core is from these cheap ferrite rod
Ferrite rod for 1MHz radio may not work as expected at 5KHz. How did you verify the values?
Hi Paul, it's nice to have you here. I'm using this Chinese guy for all of my measurements:
http://obd.tui21.com/img/tui21ecutool/victor-88c-3-1-2-digital-multimeter2013311221911734.jpgI couldn't find any other type of core here except toroidal ferrite cores, that I tought they will probably saturate in high current levels. I have no idea about characteristics of these cores; I just wound 220 and 530 nuber of turns with 0.7mm enamel wire to measure 1mH and 12.5mH with the meter. I also have no idea about the meter and how it measures the inductance (I mean measurement frequency,...).
The DC resistance is 0.3R for L1 and 0.9R for L2, so as you have said the series resistance for the resonance circuit would be enough due to L2's resistance. For L1 which idea is better? Should I rewind it around an iron bolt or just use the series resistor?
Thanks again, and sorry for asking too many questions...
Sajad
> For L1
Whatever I said. 20 Ohms *parallel* to L1.
Quote from: PRR on February 18, 2016, 08:45:20 PM
> For L1
Whatever I said. 20 Ohms *parallel* to L1.
Thanks, I'll do it ASAP and I'm gonna perform another test and report the results...
Thanks again
Sajad
I put a 22R/10W resistor in parallel with L1 and here is the new impedance gragh:https://db.tt/6AvLkgAv
Every thing is almost the same except that the resistor brought the high end impedance down as I wanted, and also the lowest impedance raised a bit and got more close to the speaker, perfect!
I also tapped the output and fed the signal to a 100k resistor and then to a 20K pot for a line out; then I used my cheap zoom multieffect as headphone amp. WOW! :o Now my 18 Watts Marshall is on max volume and I'm paying it through my headphones without any significant tone loss! This thing really rocks!
Thank you Paul and thanks to all for helping me...
Sajad
what are the specs for the inductor you used?
i have been thinking about making one for myself
Quote from: Vitrolin on February 20, 2016, 02:01:02 PM
what are the specs for the inductor you used?
i have been thinking about making one for myself
I just found a ferrite rod core and wound a couple of wires around it and calculated its AL; then I calculated the number of turns using this:
http://coil32.net/ferrite-rod-core-coil.htmlThen I used 0.7mm enamel wire and wound the inductors.
Unfortunately I don't have a scope to see if the inductors get into saturation or not; but for someone who have access to different types of inductors I think it's a better idea to use some kind of inductor silmilar to air core inductors; especially if you want to use it with a 50 or 100 watts amp.
QuoteI just found a ferrite rod core and wound a couple of wires around it and calculated its AL; then I calculated the number of turns using this:http://coil32.net/ferrite-rod-core-coil.html
The formulas for rod-core inductors aren't very precise. Putting a few turns is a good start but the final AL depends on the coil length (as the site indicates) and also the build-up of turns . So you need to measure the inductance after you have finished and tweak the turns.
BTW, that site is pretty good. There used to be another site with more/different calculations.
Regarding saturation: Ferrite saturates at quite low flux densities but fortunately when you used an open core like a rod it doesn't saturate that easily. They also saturate in a softer way compared to gapped cores due to the uneven flux density along the length of the core. It is possible to estimate the saturation current but it is a little messy. One way to help reduce the chances of saturation is to use a larger area (bigger diameter) core ferrite, or even two or three rods bundled together.
Ferrites work fine at low frequencies. Many hifi speakers manufactures have using rod cored inductors for years as part of the crossover network.
Quote from: Rob Strand on February 21, 2016, 03:36:44 AM
The formulas for rod-core inductors aren't very precise. Putting a few turns is a good start but the final AL depends on the coil length (as the site indicates) and also the build-up of turns . So you need to measure the inductance after you have finished and tweak the turns.
BTW, that site is pretty good. There used to be another site with more/different calculations.
Regarding saturation: Ferrite saturates at quite low flux densities but fortunately when you used an open core like a rod it doesn't saturate that easily. They also saturate in a softer way compared to gapped cores due to the uneven flux density along the length of the core. It is possible to estimate the saturation current but it is a little messy. One way to help reduce the chances of saturation is to use a larger area (bigger diameter) core ferrite, or even two or three rods bundled together.
Ferrites work fine at low frequencies. Many hifi speakers manufactures have using rod cored inductors for years as part of the crossover network.
Thanks, good information. I'm using my multimeter for measuring inductance, and it seems the inductor values are the same as the schematic.
The loudest amp that I already have around is an 18W; and with this amp I don't know if the inductors saturate or not... btw; Is there anyway that I could check them without a scope?
The a few ways to do it. Some methods use sinusoids and some use pulses.
Roughly:
Drive a low-level sinusoid into the inductor then calculate Z = v/i. L = Z/(2*pi*f); should agree with your meter.
Then crank-up the test current until the measured inductance falls to 90% or 70% of the low current value.
Choose f such that Z is about 10 times or more the series resistance of the inductor.
The peak current is a rough estimate of the max current.
Here's a relatively simple way for estimate the maximum current using maths:
Imax = N * A * Bsat / L
where
N = number of turns,
L = inductance in H,
Bsat is saturation flux density in T,
A = area of the core in [m^2]
For your inductors:
Bsat [T] 0.3 0.3 ; determined by ferrite material
N 220 530
d_rod [m] 1.00E-002 1.00E-002
A [m^2] 7.85E-005 7.85E-005
L [H] 1.00E-003 1.25E-002
AL [H/t^2] 2.07E-008 4.45E-008
Imax_pk [A] 5.18 1.00
So I get 5.2A max for the 1mH and 1.0A for the 12.5mH
The next tricky thing is to work out the current that passes through the 12.5mH inductor when using a guitar signal.
Current will be will be less than a full-power sinusoid - which means you can use an inductor with a lower Imax.
Quote from: Rob Strand on February 21, 2016, 06:52:08 AM
Here's a relatively simple way for estimate the maximum current using maths:
Imax = N * A * Bsat / L
where
N = number of turns,
L = inductance in H,
Bsat is saturation flux density in T,
A = area of the core in [m^2]
For your inductors:
Bsat [T] 0.3 0.3 ; determined by ferrite material
N 220 530
d_rod [m] 1.00E-002 1.00E-002
A [m^2] 7.85E-005 7.85E-005
L [H] 1.00E-003 1.25E-002
AL [H/t^2] 2.07E-008 4.45E-008
Imax_pk [A] 5.18 1.00
So I get 5.2A max for the 1mH and 1.0A for the 12.5mH
Thanks man, your informations just made my day.
So if we want to assume the worst case scenario in the case of 18Watter;
Pmax(square)= 36W into 8 ohms load
Vpeak= (Pmax*Rload)^0.5 = 17V
Ipeak= Psquare/Vpeak = 2.12A
So I guess the 1mH should be fine, but the 12.5mH is getting into saturation at max volume :icon_neutral:
Unfortunately I can't find any other type of core around here; so can I just glue let's say 4 of these cores with each other and wind another 12.5m inductor with more current capability? What's your idea?
Thanks again
Sajad
EDIT: I calculated the magnetic permeability(u) of these cores to be around 3850. So by increasing the core aread(a) by means of putting two or three cores in parallel, 'Al' becomes larger and subsequently 'N' gets around 177. According to the saturation current formula this won't increase the current capability very much, am I right?
Comparative FR curves for 4x12 cab & RLC loadbox (output of PA with 10-15 Ohm output impedance)
(http://milas.spb.ru/~kmg/files/projects/LoadBox/Cab_Load.jpg)
Loadbox schematic (very similar to powerbrake RLC part but invented independently)
(http://milas.spb.ru/~kmg/files/projects/LoadBox/Load.jpg)
Want to say some precautions about bass resonance inductor & capacior. Current through inductor & capacior at resonance frequency is about 2.5 times higher then current from PA output. So if we have 3.5A from PA 100W/8Ohm, current through inductor & capacior at resonance frequency will be about 9A.
Quote from: KMG on February 21, 2016, 02:59:04 PM
Comparative FR curves for 4x12 cab & RLC loadbox (output of PA with 10-15 Ohm output impedance)
(http://milas.spb.ru/~kmg/files/projects/LoadBox/Cab_Load.jpg)
Loadbox schematic (very similar to powerbrake RLC part but invented independently)
(http://milas.spb.ru/~kmg/files/projects/LoadBox/Load.jpg)
Want to say some precautions about bass resonance inductor & capacior. Current through inductor & capacior at resonance frequency is about 2.5 times higher then current from PA output. So if we have 3.5A from PA 100W/8Ohm, current through inductor & capacior at resonance frequency will be about 9A.
Thanks KMG, I'm also a fan of your FET designs.
Can I replace LRC part of the power brake with your design?(I'm planning to use it as an attenuator mostly!)
Also what kind of inductor core material you suggest for those inductors?
Quote from: sajy_ho on February 21, 2016, 12:17:53 AM
Quote from: Vitrolin on February 20, 2016, 02:01:02 PM
what are the specs for the inductor you used?
i have been thinking about making one for myself
I just found a ferrite rod core and wound a couple of wires around it and calculated its AL; then I calculated the number of turns using this:http://coil32.net/ferrite-rod-core-coil.html
Then I used 0.7mm enamel wire and wound the inductors.
Unfortunately I don't have a scope to see if the inductors get into saturation or not; but for someone who have access to different types of inductors I think it's a better idea to use some kind of inductor silmilar to air core inductors; especially if you want to use it with a 50 or 100 watts amp.
i wanted to know about the inductor with many taps used as attenuator
> Can I replace LRC part of the power brake with your design?
They are "the same".
It is a 3-part series circuit, the parts can be in any order.
There are small value differences, typical of differences between size or model of speaker.
The Marshall omits a treble-damp resistor. It mostly works above the guitar band. Also omitting it makes the sound a bit brighter, and "stuff" between the amp and the speaker tends to dull the sound (or you think it does), so a zingy top end is good marketing.
(http://oi66.tinypic.com/fcmqg4.jpg)
So I guess I better to stick with the current arrangement; but can someone tell me which type of core is suitable for the inductors to increase their saturation current? I mean I know that the best idea would be air core for the 1mH inductor, but what about the other one?
Thanks
Quote from: Vitrolin on February 21, 2016, 06:38:16 PM
i wanted to know about the inductor with many taps used as attenuator
You mean the tapped inductor in the right side of the schematic?
It's an autotransformer, you can use any type of tapped autotransformer. For that I used a 100V single winding speaker matching transformer that I found in my parts.
QuoteIpeak= Psquare/Vpeak = 2.12A
I calculated the current through the 12.5mH inductor alone using a circuit simulator and got the same sort of value at the worst-case frequency.
Quote
EDIT: I calculated the magnetic permeability(u) of these cores to be around 3850. So by increasing the core aread(a) by means of putting two or three cores in parallel, 'Al' becomes larger and subsequently 'N' gets around 177. According to the saturation current formula this won't increase the current capability very much, am I right?
I think the problem is the formulas say to use the magnetic permeability of the core (ur)
but I seem to remember you should use the "rod permeability" urod. FYI rods often use low ur material.
The rod permeability is a function of the rod length to rod diameter ratio and is usually given as a graph:
http://www.magneticsgroup.com/pdf/erods.pdf
Again, from what I remember the AL factor doesn't change much and probably gets a little larger.
When you use n cores:
use the total area A total = n times A for one core, and,
for the diameter in the urod calculation you will need to use an effective value
d_effective = sqrt(n) * d for one core.
You might scrape by with two rods but three is safer from a max current perspective and is easier to bundle.
Edit: If AL is about the same N is about the same and since A is n times larger the maximum current should be n times larger.
Watch out playing around with inductor losses. What seems better overall to the eye might not sound better. The mismatch at low frequencies that occurs changes the "presence". Hard to know what is better without doing AB tests.
Quote from: Rob Strand on February 22, 2016, 05:49:23 AM
QuoteIpeak= Psquare/Vpeak = 2.12A
I calculated the current through the 12.5mH inductor alone using a circuit simulator and got the same sort of value at the worst-case frequency.
Quote
EDIT: I calculated the magnetic permeability(u) of these cores to be around 3850. So by increasing the core aread(a) by means of putting two or three cores in parallel, 'Al' becomes larger and subsequently 'N' gets around 177. According to the saturation current formula this won't increase the current capability very much, am I right?
I think the problem is the formulas say to use the magnetic permeability of the core (ur)
but I seem to remember you should use the "rod permeability" urod. FYI rods often use low ur material.
The rod permeability is a function of the rod length to rod diameter ratio and is usually given as a graph:
http://www.magneticsgroup.com/pdf/erods.pdf
Again, from what I remember the AL factor doesn't change much and probably gets a little larger.
When you use n cores:
use the total area A total = n times A for one core, and,
for the diameter in the urod calculation you will need to use an effective value
d_effective = sqrt(n) * d for one core.
You might scrape by with two rods but three is safer from a max current perspective and is easier to bundle.
Edit: If AL is about the same N is about the same and since A is n times larger the maximum current should be n times larger.
You're right, I think I misunderstood the core permeabilty with the MTERIAL permeability. So I guess I need to go to store and buy 3 more cores.
Thanks again for all your help...
Sajad
No worries - this stuff can get confusing.
FYI, here's another spin of the inductance calculations:
http://agata.pd.infn.it/LLP_Carrier/New_ATCA_Carrier_web/DataSheets/New_Symbols/15th_edition_Fair-Rite_catalog.pdf\
see page 98.
Quote from: Rob Strand on February 22, 2016, 06:17:31 AM
No worries - this stuff can get confusing.
FYI, here's another spin of the inductance calculations:
http://agata.pd.infn.it/LLP_Carrier/New_ATCA_Carrier_web/DataSheets/New_Symbols/15th_edition_Fair-Rite_catalog.pdf\
see page 98.
Thanks again Rob.
I wound the 1mH inductor again just for experiment, andThis time I went air core, it was my first air core inductor and I guess it'll be the last one :icon_lol:
The inductance value turned out very good and the DC resistance is only 0.5R; the only problem is that some windings are so messy; so I guess the leakage inductance will be great. What would be the possible consequences?
> the DC resistance is only 0.5R
Use smaller wire.
The treble inductance is in-series with the ~~8 Ohm "voice coil resistance". You could use wire 1/4 diameter (1/16 area) and get the DCR up to 8 Ohms; then you do not need the resistor! However heat will be an issue (the wire can't run as hot as a resistor, the coil may have less surface area to throw heat). But it would be quite valid to wind for 1 or 2 Ohms in the coil, take the difference out of the "8 Ohm" resistor.
> some windings are so messy; so I guess the leakage inductance will be great.
Leakage inductance is mostly about transformers, not single-winding coils.
Also where leakage inductance "might" be a problem, the impedances are transitioning to the 20 Ohm resistor across the coil.
An air-core coil is best wound on a form. Old plastic solder-spool. Or a wood stick and two heavy cardboard flanges. For this purpose, you don't have to wind in neat layers, just don't get so bunched-up that wire pinches itself and breaks the enamel. (Even then, this would be "OK" because it approximates the losses of a real voice-coil in a conductive magnet gap; but we don't know how much is OK, and "pinch" is not predictable or consistent over time.)
Quote from: PRR on February 22, 2016, 01:14:22 PM
Leakage inductance is mostly about transformers, not single-winding coils.
Also where leakage inductance "might" be a problem, the impedances are transitioning to the 20 Ohm resistor across the coil.
Thanks again Paul, here's the mess that I'm talking about:(https://db.tt/G6X25Zmq)
So if there isn't any problem with being messy, this inductor would be good in series with 8.2R resistor.
The only thing remaining is the resonance inductor; that I probably follow Rob's idea and stack 3 or 4 rod cores together and wind another one.
Thank you all guys...
My up to 15W loadbox
(http://milas.spb.ru/~kmg/files/projects/LoadBox/brdtop.jpg)
Very interesting idea about creating transformerless attenuator with equal In/out FR
(http://s2.postimg.org/s6w9a6v21/fig3.gif)
Only limitation to keep In/out FR:
P(out)=K*P(in) @ K<= 0.16.
Original article (unfortunately in Russian language)
http://forum.amtelectronics.com/index.php?topic=33.0
> attenuator with equal In/out FR
Neat.
Note that output impedance is not emulated (that is very difficult).
A basic SS amp is zero Ohms out.
Classic Fender (5F6a) is (for 8 Ohm connection) 15-20 Ohms output impedance.
This plan is near 4 Ohms (on any amp).
The reaction of driving impedance on the loudspeaker in use affects its response and damping.
However this plan does give good power-loss and response quite simply.