EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: JuiceKing on April 15, 2012, 02:43:07 am
-
Hi,
I'm interested in exploring the effects of various components on sweep signals. I'm using an iPad app (SignalScope 3.0.4 Signal Generator by faberacoustical) as a sweep generator and an oscilloscope to look at it. The sweep takes about a second to cover a 0-20kHz range. According to the SignalScope doc, the log sweep (which I am using) exhibits a pink magnitude response--i.e., a 3db/oct. drop in magnitude with increasing frequency.
On the oscilloscope, I see the attached trace. This is a single sweep photographed with a second exposure. The trigger is set so that the scope sweep starts with the signal generator frequency sweep, so the low frequency signal is on the left and the wavelength gets progressively shorter as you move to the right.
I'm puzzled, though, why the overall waveform rises as frequency rises. I can't understand where that rising DC offset would come from. Also, does the tapering look right to you--3 db/oct. seems to take a long time to kick in and then it does with a vengeance at the end...is that to be expected with a logarithmic sweep?
Thanks for any help you can offer!
- Ken
-
Re the DC shift:
You probably need to be AC coupled.
It probably shouldn't happen.but I've seen similar effects, with video waveforms which suggest that the gen is DC coupled
like a video gen.
If the sweep is logarithmic,the change in frequency per division will become greater as the display progresses .
The response thing is a bit confusing.It seems strange that you would want to test anything with a tapering response from the source.
Usually signal generators attempt to maintain the output level within 1dB ,over the useable range.
With "pretend" signal generators,anything can happen though! ;D
-
I thought I should add an explanatory note about my comment on video signals.
My original reply could be confusing.
A sweep signal used on an analog TV system usually occupies two fields.
This means that the the DC component is actually two fields of lines consisting of that DC level,with attached syncs.
Circuits respond to this as a long rectangular pulse at 25Hz.(for the PAL system)
If the DUT is a Band Pass Filter (BPF) with a low frequency limit higher than this,or it is capacitively coupled with too small a value of capacitor,
The DC centreline of the sweep will appear to have DC shift from one end to another,if observed with a DC coupled Oscilloscope.
The normal "fix" if you have to look at such a device with a video sweep,is to use the 'scopes AC coupling function.
This will remove the DC shift.It isn't always perfect,but is usually good enough.
The DC shift in your picture looks very similar,so possibly your generator App produces a slow sweep with a DC component.
-
Hi vk6zgo,
Thanks so much for your reply and info. I should have mentioned that I tried to filter out the DC component by switching the AC coupling, but it didn't help. My original post misstated the sweep speed--it's roughly 10/sec, not 1/sec, so the rise isn't exactly DC anymore!
I also looked at the "linear" mode sweep. Oddly, that one tapers as well at the high frequency-end of the sweep, and has various amplitude anomalies across the band as well.
The SignalScope app was not one of your typical "$0.99" wonders...it was part of suite of apps that cost $40-50...the single most expensive app I bought on the iPad, so I was hoping for more than a toy...too bad. It SHOULD work...or be explainable, at least. Its synthesized waves are clearly challenged by bandwidth limitations. Take a look at the 1kHz square wave... Sine waves are smooth at 20kHz, but there's a fair amount of high frequency hash that needs to be filtered for triggering.
I will check in with the SignalScope forum/vendor to see if there are any ideas there. The doc says that the range of the sweep is adjustable but in my experience it's not--it does the full sweep of frequencies no matter what I try.
Thanks again for your help!
- Ken
-
I would say the poor signal is as a result of the source. The audio stages of a phone are not exactly hifi, or even, in many cases, linear or even close to it. They generally are bandwidth shaped to a phone line quality, with possibly some added equalisation for the speaker frequency response ( or the included headphones).
Try to terminate the output of the phone with 47 ohms ( close to the expected headphone resistance) and try again, with a little more gain on the scope input, and without using the X10 probe, rather a X1 probe, with this low impedance. This should improve the output if the problem is the output stage being non linear into an open or high impedance.
-
Hi Sean,
Thanks for your suggestion. I just tried adding a 47-ohm resistor in parallel to the output--no change. The resistance of my headphones is 16-ohms, so I tried an 18-ohm resistor--same result. I do find that on a pure sine wave, the 18-ohm terminating resistor does reduce ultrasonic frequency hash--the sine wave on the scope is a little cleaner/sharper.
At least for settable sine wave output, from 20Hz - 20kHz, output is flat--I'm seeing less than 1% change in RMS voltage...so I think that the iPad hardware and OS can output reliably at least for this function. I'm starting the suspect that the SignalScope software doesn't do what it's supposed to do in sweep mode. I'm waiting for permission to post on their forum. At their website, I'm reminded that this software cost $75(!) so I'm hoping that it works or there's someone who will fix bugs if it turns out there are any!
- Ken